Yes and no.

If the goal is to have the best support for all future games, sure.

But that's not what protons main goal is.

The main goal is to get all the hundreds of thousands of games that have already been made, to work.

Making one game for linux, only produces one game.

Getting proton right, gets every past game that has ever worked on windows, to work on linux.

That a lot of future games will also work thanks to this, and hence require less work to run on linux, is a bonus. Not the main goal.

Hence, proton is ABSOLUTELY worth it. Even if microsoft released directx 13 tomorrow, that wouldn't break any of the games that already work. The alternative is to go back and port each game one at a time. Proton isn't just happening instead of linux ports. It's also happening instead of porting all old games, and enables running games with shuttered studios, unclear IP onwership, lost source code or lost interest.

If you REALLY want appimages for archiving, the best way would be to package an individual game, a proton wrapper with all the dependencies, and compile that up into a neat appimage. Then you'd have a "future proof" executable file for that one game to keep and run in the future.

But that's ignoring the fact that appimages and flatpaks aren't truly that agnostic. Changes in graphics APIs happen on linux, too.

And it makes way more sense to use the same proton for multiple games, instead of bundling an extra copy of it with each game.