Replace spirv_std::storage_class::X<T> with T/&mut T and #[spirv(x)] &T/&mut T.

[eddyb]

This is my pre-#416 branch, I started this as part of #414, but only later realized I don't need it.

It's easier than I expected because, once again, I was able to mostly reuse existing logic (attribute parsing in this case).

As it's currently implemented, rules are:

• only &T and &mut T are allowed as entry parameter types, with explicit storage class attributes
  • should probably be &mut MaybeUninitialized<T> instead of &mut T, but ideally we can use -> T instead for Output
  • it considers &T immutable but it should also check that T: Freeze
• T defaults to Input, &mut T defaults to Output, and those storage classes cannot be explicitly specified
• all other storage classes use a #[spirv(...)] attribute on the parameter, e.g. this line in sky-shader:

      #[spirv(push_constant)] constants: &ShaderConstants,

• UniformConstant, and PushContant storage classes only allow &T (i.e. not &mut T) types
• Private, Function and Generic storage classes cannot be specified at all
  • should probably restrict this further based on shader stage or kernel vs shader, but we can do that later

This works and passes tests, but I'm opening it as a draft because it should wait on a decision regarding #416, especially around the Output storage class and whether we should use a -> style, support bundling them in structs etc.

EDIT: updated to reflect PR changes, i.e. Input not taking references anymore.

[khyperia]

Looks hecking great! @hrydgard and I had a conversation in discord, path forward seems to be to make inputs be by-value, and then in a follow-up, do the annoyingly complex work of making outputs be returned via tuple or struct or whatever.

[eddyb]

@khyperia FWIW this is ready for a re-review, but I can't figure out what's wrong with the CI.
I don't want to tell it to retry because it might auto-merge because of the previous review.

[XAMPPRocky]

I'm going to re-run and temporarily disable auto-merge

[XAMPPRocky]

Yeah, it was an issue with GitHub Actions, the PR passes.

[XAMPPRocky]

Would there ever be a need to use these attributes in spirv-std? If not, I agree that we should remove them.