Added sg_dealloc_xxx() and sg_uninit_xxx() suite of functions

[oviano]

These are the equivalent to the sg_alloc/init functions and allow async destruction of resources. This is useful when using sokol in a multithreaded scenario; the render thread does the sg_init/sg_uninit while a separate resource/api/update thread can call the sg_alloc/sg_dealloc.

[floooh]

Looks good! I'll try to do a "merge day" soon where I'll focus on pending PRs.

[floooh]

I'm starting to look into this PR now. I think there's two small problem:

• the uninit-functions should put the resource slot back into the SG_RESOURCESTATE_ALLOC state
• ...and the dealloc-functions need to check whether the resource slot in in the SG_RESOURCESTATE_ALLOC state

I think those changes are trivial, and I'll try to do them during the merge.

[oviano]

Yes, you are right of course, I remember wondering about that when I did it, and I think the only reason I didn't at the time was that in all cases the uninit should be followed by a dealloc anyway so it seemed like it wasn't too critical for my testing.....and then I forgot to go back to it :)

[floooh]

Ok merged, many thanks for the PR!

As described above, I've added the correct "state unwinding". I've also added a couple new tests to sokol-gfx-test.c in the sokol-samples repository:

https://github.com/floooh/sokol-samples/blob/4cf2a1ef723207e6d7a9b1500f79eb048a2f4ce6/tests/sokol-gfx-test.c#L674-L742

Another minor thing you missed is adding new entries to the sg_trace_hooks struct (this only shows up when compiling with SOKOL_TRACE_HOOKS enabled).

[oviano]

Great, thank you and also if there is a next time I'll be sure to look out for the sg_trace_hooks!