When this happens in X11, the server restarts, I see the screen go black for an instant, that's it.
When this happens in Wayland, all my open programs are killed. No save prompts or anything, just insta-killed. This makes Wayland unusable unless they can guarantee no crashes, ever.
Unfortunately, nobody seems interested in such user reports or quality control (apparently, turns out there are many like us who're getting "upgrade your hardware", "works for me", "user error", etc. I retreated to Debian stable, which is indeed very stable.
What you see might be a GPU hang or reset that recovered.
Some Wayland compositors have a window manager model by using an external process for window positioning logic (e.g. river). This is mostly just for convenience of development though, making it easy to experiment with layout engines.
Since the window manager is also the compositor in many X setups (e.g. GNOME and KDE), the WM is very much affected by GPU drivers. It doesn't have to be the GPU drivers though, plenty of complexity in the WM/compositor itself.
The core complaint is absolutely valid - the protocol and libraries should have been designed so that applications can recover from a server crash/restart, even if the legacy X server does not provide that guarantee.
The difference is that X survives a window manager crash, whereas Wayland typically does not, because on Wayland it's usually the same process as the server.
Wayland has a myriad of these sharp-edge cases which i also find each time i try it out. what worries me a little is that some of them are clearly design decisions with no simple remidiation, your crashing example (which i have also witnessed first hand) is one of them.
* https://www.youtube.com/watch?v=fRdnRwPBFBk
* https://invent.kde.org/plasma/kwin/-/wikis/Restarting
Though I'm not sure what's the current status of that is or how actively it's progressing. Hopefully it will turn into something usable, it's important.
Your windows manager crashed. If xorg had crashed you'd have lost your data.