Initial naga changes for mesh shaders

[SupaMaggie70Incorporated]

Connections
#7197

Description
This is the initial change for naga mesh shaders. Currently not in a workable state. Issues to work out:

• Expecting u32 for certain functions, right now you have to explicitly say 0u, etc
• Actually interpreting mesh shader related things in the IR
• Validation of shaders
• Sending back more information about what limits it needs
• Reflection info
• SPIR-V writing
• Testing of SPIR-V output

Testing
There is a test WGSL shader that is parsed and then written to SPIR-V

Squash or Rebase?
Absolutely squash

Checklist

• Run cargo fmt.
• Run taplo format.
• Run cargo clippy --tests. 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.

[SupaMaggie70Incorporated]

I'm marking this as ready for review so that it can generate some eyes to look at the general process. Obviously the PR is 5k lines long so will not be merged in anything like its current state. It also is failing CI due to rspirv not being updated (see #7945 ) and the shader compiling benchmark not respecting the configuration tomls and therefore trying to compile for incompatible backends like GLSL.

The fragment shader input on the test would be invalid to use with the corresponding mesh shader because it doesn't have the not yet implemented @per_primitive attribute for its primitive input.

I'd mainly like for people to check out my SPIRV backend changes, as that is the part I am most unsure about.

The generated mesh and task shaders pass spirv-val, but I haven't yet tried using them in a real mesh pipeline.

[SupaMaggie70Incorporated]

Unmarking as draft (again lol) because I have given it another pass with my own eyes and @cwfitzgerald mentioned he would check it out

Also I'd like to know how you'd prefer to address the benchmark issue

[jimblandy]

@SupaMaggie70Incorporated Would you like to rebase this and address the CI failures?

[SupaMaggie70Incorporated]

@jimblandy Rebase onto what exactly? Also, the CI failures are due to shader compiling benchmarks that can't be told "only compile this into SPIRV" since not all output languages are supported yet. So I am actually looking for guidance on how I should address that!

This is marked as a non-draft despite its size because @cwfitzgerald mentioned he wanted to take a look over the whole thing before I split it up and let other people review individual PRs. The PR isn't quite done as is, but I just wanted to give him the chance to look over approximately what it would be.

[andyleiserson]

Also, the CI failures are due to shader compiling benchmarks that can't be told "only compile this into SPIRV" since not all output languages are supported yet. So I am actually looking for guidance on how I should address that!

There are toml files alongside the test shaders that specify which backends they are intended for, although I'm not sure if what's there will match exactly with what the benchmark test needs. But an option to consider is figuring out how to reuse the code in naga/tests/naga/snapshots.rs that parses these toml files for the benchmark.