There's a valid question about how much usage patterns differ on different form factors.
It feels like the last two points you make contradict each other: there's a huge difference between "extension to desktop" and "pick up desktop stuff and go".
Obviously, for a myriad of hardware and UI reasons, we're not doing the latter typically. The people demented enough to want desktop software in their pocket buy a GPD Win or the like.
The real issue is behaviour patterns, though. Give me the performance of my desktop in a six-inch pocketable device with two day battery life, and I'm still going to use it very differently.
Where do many of us spend most of our workdays? In an IDE or office suite. No matter what you do, you're not going to be able to shrink those down and be usable for anything but the most trivial tweaks on a six inch screen with a glass keyboard. We sort of saw this attempted in the early days of Windows CE/Pocket PC with its silly little Portable Office apps. A docking station might make it usable when I get BACK to a desk, but at that point, I may as well just be using the phone as an expensive flash drive for sneakernet.
Okay, maybe gaming? Potentially, if done right. Genshin Impact making 82 bazillion dollars out of a game that scales from phones to desktops is the poster child. But as anyone with a Steam Deck can demonstrate, you have to pick stuff that was designed to work well with the form factor. That undermines the dream of "the entire universe of cool stuff your desktop has to offer".
Media consumption? Even that's different. People may prefer local storage to streaming due to unreliable/slow/expensive mobile data, and the media they consume is often different too. I don't want to binge 90-minute episodes of anything on a phone.
So we've basically tossed out most of the obvious answers for "seamless desktop extension". This means one of two things:
1) We're drilling in the wrong place. The desktop extensions we need are such paradigm shifts nobody has even given us a good demo
2) The things that define the mobile ecosystem currently-- for example, apps that lean heavily on "it has GPS and cameras", lightweight games, low-commitment consumption of media-- shape the market because they're the best use cases we've found for the form factor.
I liked Windows Phone. I still have three of them in my box of bad mobile device decisions (somewhere I think there's a Palm Pre too) What I liked about it was that it seemed to have a clear and cohesive design language, and it was reliably snappy on cheap hardware. We've basically given up on the former, but Android has managed to close the gap somewhat on the latter, probably mostly through surfing on Moore's Law.
Microsoft could have had one other chance, but I think they've already blew it: as privacy is becoming a selling point (see Apple and DuckDuckGo marketing), positioning themselves as "we have a business model that's not 96% ad-dependent, so we're not going to sell out your personal life for a quick nickel" could have made them the trustworthy choice, but they've burnt through any credibility they had on that front in the last few years.