DepthStencilState validation fixes

[andyleiserson]

The depth-related fields are optional in the WebGPU API, and must be specified iff they are relevant to the configuration. In order to have wgpu-core provide a common implementation of this validation for all the JS clients, this changes pipeline descriptors to use a new DepthStencilStateIdl struct that matches the spec (at least for the enable/compare -- it still differs from the spec in that the depth bias and stencil configuration are not inlined).

DepthStencilState is maintained as-is so that implementations can keep the same struct literals and just add a call to .into(). I also added some associated method constructors that will be forwards-compatible with making DepthStencilState look like DepthStencilStateIdl.

There is some related discussion in Matrix. We may decide to create a separate DepthState struct instead.

This PR conflicts with #8856. Whichever lands second should enable the entire depth_stencil_state suite.

Testing
Enables some relevant CTS tests.

Squash or Rebase? 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.

[andyleiserson]

(Responding to discussion of a possible DepthState struct in matrix): The rules are that depthWriteEnabled must be provided if using a depth format; depthCompare must be provided if depthWriteEnabled is true or if stencil has an active depthFailOp. So while depthWriteEnabled could be mandatory in DepthState, depthCompare would still need to be optional. I think that makes it harder to justify a breaking change to introduce DepthState.

A data point: https://github.com/search?q=language%3ARust+RenderPipelineDescriptor+DepthStencilState&type=code finds 6.7k files 😬. Even the need to add .into() doesn't seem great, but I couldn't figure out an alternative that didn't involve duplicating or adding a type parameter to all the pipeline descriptors.

[andyleiserson]

@cwfitzgerald you had proposed a different change that would create a separate DepthState struct. Does the issue of depthCompare needing to be optional even within DepthState (see #8840 (comment)) change your thinking at all?

(And I guess separately from that, there's also a question of whether to offer a migration via DepthStencilStateIdl like it is now, or to just inflict the full pain and change DepthStencilState to look the way we want.)