Allow the "gles" backend to (theoretically) compile on no_std targets.

[kpreid]

Connections
Part of #6826.

Description
Removes std usage from the gles backend in wgpu-hal, allowing wgpu-hal to build on no_std targets like wasm32v1-none (once its dependencies, notably naga, are all no_std compatible too).

This entirely consists of conditionally replacing Mutex with RefCell, then making sure that this doesn’t accidentally happen if we are also exposing Send + Sync.

I’m not very confident in the design of this change, but I tried to do it in a way which at least doesn’t add any unsound configurations.

Testing
Confirmed the backend still builds on native and wasm, and ran the test suite.

Squash or Rebase?
Rebase

Checklist

• Run cargo fmt.
• Run taplo format.
• Run cargo clippy. If applicable, add:
  • --target wasm32-unknown-unknown
• Run cargo xtask test to run tests.
• If this contains user-facing changes, add a CHANGELOG.md entry.

[kpreid]

I notice that before this change, cfg-if is only a dev-dependency, never a library dependency. I assume there is no strong reason for that; I can rewrite the code to use only #[cfg] if desired.

[Wumpf]

it's an absolutely tiny dependency and makes code look nicer here, don't mind it :)

[kpreid]

Ah, I see I broke Emscripten. It’s late, so I will think about that another day.

[kpreid]

There we go. I just forgot that cfg(target = "...") isn’t a thing.

[Wumpf]

looks good to me. Changes are non-invasive enough imho
We should follow-up though with a way to check this configuration on ci, lest someone adds e.g. a new mutex or similar again

[kpreid]

A sufficient CI test would be

cargo (check|build|clippy) -p wgpu-hal --features gles

but once #6826 is closer to done, we can and should run builds on actual doesn’t-have-std targets, which will be even stronger.