Let me be the first to tell you that for many, many people, smartphones are their primary computing platforms, and they do everything on them. I've had to help people file taxes on smart phones, sign documents, and more. Processor speed isn't important now so much as it is later. A good portion of the budget market is going to grab a phone because of the price and hold on to it for years. Will a 20% slower processor stand up as well as its bigger cousin in the future? Probably less so.
Normal users need good performance too, but the bottom end of the market is always saturated with completely unsuitable crap. A safe way to look at it is that most users need roughly the power of a 2 year old flagship device. That has been true for many years as well.
It absolutely is getting more correct. I used to update my desktop every year or two. I'm now still on a Core i5 2500 from 2011 and have zero reason to upgrade again, even despite some gaming use.
For phones, it's a bit different because they're back on the part of the performance curve where desktops were years ago and power efficiency is still improving rapidly but I think they're getting close to that crossover point.
The average user can comfortably use a 5 year old pc even today, ofcourse with an SSD upgrade.
For me, I'm planning on getting a Pixel 3a. The talk of it being like a "2 year old flagship" sounds great to me, not a detriment at all. I'm upgrading from a Honor 8, which is an almost 3 year old mid-range phone, so the Pixel 3a will be a decent upgrade, even though the performance of my Honor 8 is still perfectly good. I'm mainly upgrading for the camera, 3 years of software updates and because my Honor 8's Bluetooth is becoming less reliable over time.
The limiting factor of a cellphone as a primary computer platform is actually lack of keyboard and screen space. Computing power for most people has been fine for years now.
By far the heaviest app I use on my phone is the browser. Not the browser per se but some of the websites I visit, especially ecommerce stores.
Apple already is doing in-house development; that's why they poached the head of Intel's in-house wireless modem chip team. But they clearly don't have a product yet, or Apple wouldn't have paid several billion dollars to Qualcomm to settle up.
Not sure why you are getting that idea. But that is not true. Even including cases where Qualcomm Adreno 640 used in 855 under some overclocked condition.
And unless something truly magical happens I don't see Qualcomm will catch up, GPU performance scales linearly with Die Size used for GPU, and Apple will forever have the advantage in that area, where Apple are using the SoC themselves and could afford larger die, while Qualcomm makes profits on die and yield of their SoC.