"Alternatively it might make sense to build some type of bespoke solution on top of a specific wayland stack, like re implementing what you get of talon in a kde plugin or via sway IPC. This seems viable to me but an incredible amount of work."
I think that is the only way forward. There is no "Linux desktop". There is KDE, Gnome etc. and if you want to do "system utilities" you have to target one of those.
The point though is that there used to be a «Linux desktop» you could target before the Wayland transition. Fragmentation of an already small market segment is unfortunate.
Sure. I use KDE myself. I actually like the Gnome default UI/UX more, but I'm really not fond of their "our way or the highway" approach when it comes to customization and interoperability. KDE feels more in line with the true FOSS "spirit", and after a bit of customization, fits my needs better than Gnome.
Huh? Fcitx and IBus both worked on GNOME and KDE as far as I'm aware. Now, Fcitx using QT and IBus using GTK helps them feel more native on KDE and GNOME respectively, but they would both work.
It seems to me that the issues on that page are Wayland-specific; anectotally, on my random X window manager works fine with Fcitx (except for Emacs, but that's probably Emacs' fault, not the IME protocol's).
No, that's a terrible idea. The right thing to do is design a Wayland protocol that gives you the access you need, get it accepted into wayland-protocols, and wait for all the compositors to implement it.
Yes, that's a slow, annoying process, but doing something bespoke means either you do the same work over and over for every compositor, or you only support one or two compositors. Neither of those is a good result.
I think that is the only way forward. There is no "Linux desktop". There is KDE, Gnome etc. and if you want to do "system utilities" you have to target one of those.