Captain's Log: Stardate 77425.6

Yesterday I finished cleaning up the GPU warp alignment stuff, getting it well-tested, and now I'm pretty happy with it overall. I added some basic support for injecting random MIDI events into the GPU simulation fuzz tester, which is primitive for now but I'm happy to finally have that since it gives more confidence that it's hard to crash Anukari.

I also made some long-delayed improvements to the UX for limiting the number of entities the user can create based on the GPU constraints. This is fairly tricky stuff; different entities require different resources (mostly different kinds of GPU memory), and then of course they all share some of these resources. So there are independent limits on the number of, say, masses and microphones, but also if you create enough microphones then that will eventually limit the number of masses in a dependent way.

The independent limits have worked for a long time. If you try to add too many of some kind of entity, there's UX to tell you that you can't. And if you load a preset that's too big for your machine, it will load it in a sort of "detached" state, and tell you what you need to do to make it fit (e.g. remove N masses, etc).

But now, the dependent limits work and are communicated to the user. This is cool because now if you want to create the absolute maximum number of masses, you can do that, it just cuts into how many microphones you can create, etc. So the user has complete control, even if some of the options are weird.