why it would be "extremely difficult with m4/3 and other small-sized equivalents" ?
1) m4/3 and similar cameras use EVFs (electronic viewfinders) instead of optical ones - these have inherent and percievable lag. It's getting better, but still present (and due to the way they work, it's unlikely to be eliminated in the near future). This means that what you see through the viewfinder is a bit behind what's really happening, making precise timing hard.
2) Focussing on fast-moving targets - espcially if the background is "busy" (i.e. has lots of contrast) is very hard for the camera (and user!) to get right. Most non-DSLR cameras use contrast-based autofocus, which is more accurate for still scenes, but slower and worse at dealing with demanding situations like the one I illustrated. DSLRs use phase detection sensors, which are much better at highly demanding situations like these (although slightly less accurate with still subjects).
Neither of these means that it's impossible - but it does mean that in situations like the one I described your success rate - shots that are in focus, and where the timing is correct - tends to be significantly lower than with a highend DSLR. When you're trying to capture a rare event (again like the one I described) that you only have one attempt at per day/week/whatever, a significantly reduced success rate is a really big problem.
Another reason is that the EVF is a screen with all its inherent issues. A screen is its own light source, compared to a DSLR finder which is the sunlight bouncing off the scene via the mirror directly into your eye. Your eye via a mirror can see a much larger dynamic range than the EVF screen, and you can make a call on exposure based on that. That said, if the EVF was truly representative of what the sensor was picking up, it may actually be useful, but the EVF is almost certainly showing a quantized version of what the sensor is seeing, and when you get the image in Lightroom or whatever it'll be different to what you saw on the EVF.
1) One dimensional phase detect sensors, rather than the (more expensive) cross-hatch ones used in DSLRs, which don't perform as well with movement or low light conditions.
2) Reduced processing power, lack of dedicated autofocus processing silicon, and less sophisicated data pipeline in general.
3) Greatly reduced user control. DSLRs, pro ones especially, have enormously sophisiticated control systems, and the ability to completely switch setup very quickly. There are a huge number of ways I can influence what descions the autofocus system makes on my cameras, and I use them to make sure the camera corretly interprets what part of what object it should be focussing on. I probably want the camera to be focussing on the head of the bird (rather than the wingtip, which is typically what a m4/3 system will go for), but that's unlikely to be in the exact centre of the frame, and definitely isn't the closest or most distinct object in the scene, so I need to tell the camera (very quickly and accurately) what it should do. Small mirrorless cameras with limited physical buttons and less tightly controllable focus systems make this very hard.
Highend DSLRs can also beat that continuous shooting rate - a Nikon D5, for example, will do 12fps (or 14fps with the mirror locked up). However frames per second isn't always relevent - you're going to be at 1/2000s minimum for this sort of shot, which means that even at 12fps you're capturing only ~ 0.6% of the time. If there's one specific moment you're targetting, even at very high fps, holding the shutter down is a bad strategy. It's useful in some situations (fps is important for pro sports photography for example), but it's often less important than a lot of people seem to believe.
DSLRs, pro ones especially, have enormously sophisiticated control systems... There are a huge number of ways I can influence what descions the autofocus system makes
If anything, the control system in my m43 camera is more sophisticated. And the autofocus is a big part of that. Similar to many cell phones, my m43 camera has a touch screen that allows me to tap the part of the scene I'd like it to use for focusing. This is a better system than any DSLR I've used. I doubt that in autopilot it would pick the wingtip: because it's in motion it's likely to be less sharp than, say, the head.
Regarding the lag in the EVF, it's certainly noticeable. But I don't see that as a problem in this use case. My experience with capturing moving subjects is that you don't even try to get that one single perfect shot. Instead you're going to be in high-speed shooting mode, grabbing many images of the scene over at least a second or so, in order to pick the best of the set. In this case, the frame rate really is important.
Finally, there's a real advantage to a mirrorless camera for the landscape work I usually do. A lot of the shots I'm looking for are long exposures where it's imperative for the camera to be absolutely still (and I've found that your approach shouldn't be "use a tripod when necessary", but instead "always use a tripod unless it's impractical to do so"). Even the internal operations of the camera can introduce unwanted vibrations. Most DSLRs (unfortunately my DSLR is not one of these) allow you lock up the mirror, which is the largest offender, but that precludes use of the optical viewfinder. But even farther, my mirrorless m43 also allows switching to full-electronic shutter, so there's nothing mechanical at all to shake the camera; I'm not aware of any DSLR that has such a mode, although there may be something out there.
The cameras I was really talking about are professional and semi professional tools - which right now is cameras like the canon 1dx II, nikon d5 (pro) and canon 5d III, nikon d810 (semi-pro). Pentax, I'm afraid, doesn't really have anything comparable right now. The control systems of these bodies are very, very different to what you're used to - the learning curve is far steeper, they're less forgiving of mistakes, but in good hands are remarkably effective tools. They also cost rather more - between USD $2700 and $6500 for the body alone.
Holding the release down and choosing the best image of the set is a common way to capture images - as I explained however, if there's a specific moment you want to capture you're very unlikely to get it this way. If you're taking a series of shots of something moving and you just want the best one? Well then buffer size becomes at least as important, if not more so, than framerate. It doesn't really help if you capture 11fps but can only do so for one second when the person next to you can do 9fps for 20 seconds. I'd also note that many users of high-end DSLRs have spent this money partly because they are indeed able to get that single perfect shot, due to lots and lots of practice.
You're absolutely correct about mirror induced vibration being a real problem for some shots. Electronic first curtain shutters are actually present in some highend DSLRs (the nikon d810 for example) - but as you say, mirrorless systems tend to be excellent at landscape photography (although sadly tilt/shift lenses for mirrorless cameras are few and far between). The point I was attempt to make was that DSLRs are masters of flexibility - m4/3 cameras do some things exceedingly well, and some things not quite well. DSLRs tend to be good at pretty much everything, if you can live with the size, weight and cost. A great deal of professionals have both - mirrorless systems for the times when they work well, and a DSLR for everything else.
P.s. in response to your paragraph on wingtip focus I've a couple of points. Firstly, touchscreen focus is great for many things, but it REALLY is not going to help you capture a fast moving bird in flight. Secondly, the head and the wings are both moving, in multiple directions at once. Neither are differently "sharp" to the autofocus sensor. The reason wingtip focus is such an incredibly common error with birds in flight photography is because the wingtip is closer, depending on the angle may sometimes move in front of the head, has more contrast and tends to be bigger (depending on the species). All these factors combine, and thus why it's a very common issue. Note that you'll only see it if you have a pretty shallow depth of field - which you're less likely to have on a small sensor camera, or with a slow lens.
Splitting hairs, but I don't think that's quite right. The medium-format 645D is a step above those you've cited (yet with a comparable price). And the recent K-1 (not to be confused with the K-01, nice job marketing guys) should be on par with the D810, etc. (see [1] and [2]). However, I'll grant that whatever else it might be doing with image quality, Pentax's autofocus systems have always been sub-par. They're improving, but so is the competition.
I'll defer to your experience with birds; I admit to being crappy with that. But I think in your example of capturing the ultimate instant, autofocus does not enter the picture (so to speak). If you're worried about the EVF delay, the (lack of) speed of autofocus will put you right out of the ballpark. If you want that decisive moment, you're going to need to be pre-focused. Or are you tracking the subject continuously? I have particularly bad results trying that.
The point I was attempt to make was that DSLRs are masters of flexibility ... if you can live with the size, weight and cost
And indeed, those latter issues are a big tradeoff, as the OP notes and I've found myself - since I got the Olympus I've hardly used the Pentax at all. You can't just wave them off as caveats, they really are important. If I'm willing to use the kit zoom lens, I can actually fit the thing into a pocket, which is a big part of why I bought it. As the saying goes, "the best camera is the one you have with you".
But it's not just availability, it's also handling and agility. The weight of, say, a 300mm equivalent lens on a m43 body is a small fraction of the same for full frame. That translates into better stability when forced to hand-hold, or ability to move faster onto target even when on a tripod, which is important when acquiring and tracking a moving subject.
On the other hand, that same thing is a detriment for portrait photography with a m43. The same geometry that works well for long lenses means that it's difficult to impossible to achieve the short depth of field that you can get with a bigger-sensor camera. I think Sony may have that covered, but for reasons we needn't get into here, Sony products aren't allowed in my house.
[1] DXO scores: K-1 is 96, D810 is 97 (https://www.dxomark.com/Cameras/Compare/Side-by-side/Pentax-... )
[2] "Image Quality is where the K-1 definitely outshines the majority of its competition at this price point, with the camera giving one of the best Raw dynamic range results we've ever seen, falling just short of the industry-leading Nikon D810. Low light Raw image quality is a match for just about anything on the market, falling just short of the class-leading Sony a7R II." (https://www.dpreview.com/reviews/pentax-k-1/10 )
I think in many ways we might actually be in agreement, albeit from slightly different perspectives. Two notes though:
1) Weight is actually an asset for stability, assuming your support system is sorted properly (by which I mean a rock-solid tripod with a proper wimberley head). The weight dampens vibrations, and the support system means that it's perfectly balanced, and can smoothly track moving objects, even at speed (which helps the autofocus a great deal). I don't know anyone who'd attempt to handhold a D5 with an 500mm f4 lens - that's just not going to work :)
2) DxO scores aren't something I get overly excited by. The important things about a shot are firstly the composition, which has nothing to do with the sensor at all, and after that comes making sure that the things that are supposed to be in focus are critically sharp, which comes down to focus (and good glass) - but that's difficult to assign a numerical score to so DxO doesn't do that. Instead they obsess over incredibly minor differences in colour rendition (which after it's been through the awful adobe raw converter won't matter anyway), performance at ISO levels way too high to actually use, and differences in dynamic range way above what any printer can reproduce. Not that useful, at least to me.
The availability is indeed a valid point, and is basically why I'd buy this product should it turn out to do what it says it'll do. Having said that, I carry a heavy camera bag around with me a lot of the time, and for me that's a tradeoff I'm happy to make for the flexibility of being able to take lots of different kinds of shots when I'm out. That's really what I meant by this whole thread - the product mentioned will be fantastic for capturing some landscapes when I don't have the option of a full camera bag, but it's never going to be able to replace the camera bag. Which is why it's (for me) a DSLR addition, not a replacement.
EDIT: Forgot to mention - I 100% agree about the image quality, and that's what I meant when I said lots of amazing shots are taken by people with very similar levels of equipment. The high end pro bodies tend not to give you that much more image quality (some of them give you more resolution, but again, that's something that gets overhyped, and isn't always needed) - they give you more controls, more durability, more buffer and so on, but with the right talent, technique (and glass) you can take a NatGeo worthy image on a $500 camera, no problem.
I do wonder if it might still struggle to keep up with the controllability of the pro DSLRs, which I discussed in a different reply, but until I can actually try one that's likely to remain an open question.
Basically, my experience is that the control surface of my E-M5ii is superior to that of my DSLR, especially in the area of autofocus.
It also has some cool stuff that I've never heard of a DSLR doing. For example, my camera does exposure bracketing by automatically driving the autofocus servos; the E-M1 (and in the future, E-M1ii) take that to automatic in-camera focus stacking.
I will say that every manufacturer seems to have a different approach to how their various exposure priority modes trade off. Like when I'm in Aperture priority, how much does it change Shutter vs ISO? I dislike how Olympus does that, and actually think that Pentax's heuristics are the best I've seen. My point, I guess, is that the control systems are always just something you have to learn to work with given the way you think about the camera operation.
Also, long - as in really long - lenses are hard to come by in MFT. Most useful for wildlife.
It has a weaker native lens selection than M43 but it has a 70-300mm lens that translates to 190-810mm (factoring 2.7 crop factor) that birders love. It can also use adapted F mount lenses with decently fast AF (center point only).
On the other hand, the N1 series has so many other weaknesses (like being massively overpriced for what you're getting) that it's not a great system for most people.