Backend change from Win32 menu: Add quick workaround for instance counter misbehavior

[hrydgard]

Should really do this some better way.. but this is quick and safe for now.

See #18148

[anr2me]

May be by sending WM_QUIT message using PostQuitMessage?

[unknownbrackets]

There are other problems too. On Linux, if PPSSPP crashes... you're pretty much (as a user) left with being unable to save settings in PPSSPP until you restart your computer.

Maybe as a fallback, each instance should write a timestamp periodically, and on startup it should count the number of timestamps that are more recent than 60 seconds ago or something. At least then the situation would self-resolve in edge cases. Still not perfect, though.

-[Unknown]

[hrydgard]

Hm. I haven't noticed behavior that bad on Linux, although I suppose most of my "crashes" while debugging end up as Ctrl+C, and then it's somehow not an issue? Either way, this seems worth to break out into a new issure report, so I'll do that.

[unknownbrackets]

It happened to me often enough on RISC-V testing that I temporarily hardcoded PPSSPP_ID = 1;... but I was crashing it pretty often in the beginning getting jit to work.

-[Unknown]

[hrydgard]

#18204