Update on review from my end: I plan on getting to this before the end of the week, but it's not my highest priority. Hopefully, it won't have a lot of competition. 🤞🏻

@ErichDonGubler Ok, this PR is in kinda a limbo right now anyway. It would need a couple texture format fixes to land before we actually start breaking the CI and testing on everybody's machines. Plus I think a lot of it is probably already covered by the CTS. Make of that what you will.

Glad you didn't plan on keeping any of those commits intact! 😅 It let me easily force-push a squashed history on top of trunk just now.

Aggh I really need to start using this whole "squash and force push" thing you guys are constantly doing.

Okay, so, I spent a bunch of time reviewing this, and after a mildly insane experiment, I pushed 1978d63 on top of what's already here, which uses @kpreid's suggestion to derive everything using macro_rules_attribute::derive. PTAL!

I'm gonna put this back into draft; there are definitely some issues that need to be resolved. In rough order of importance that I see:

1. Figure out why tests are failing.
2. Figure out why tests are OOM-ing, at least on my machine.
3. Make test enumeration less prone to manual error. Having to list a test per texture format sounds problematic enough that it's worth figuring out if there's another way to do it before merging, methinks.

I'm gonna put these issues in your court, but I'm happy to help with (3) and review any fix PRs that can be broken out for (1) and (2).

@ErichDonGubler I've reached out to you on Matrix. I'd also like to link #8670 as a sort of list of texture format issues I've ran into.