Add STRICT_WEBGPU_COMPLIANCE instance flag

[teoxoy]

Connections
Fixes #3274.
Fixes #9436.
Fixes #3365.

Description
Adds the STRICT_WEBGPU_COMPLIANCE flag, intended to be used by wgpu-core users that want to strictly limit the behavior of wgpu-core to the one defined by the WebGPU spec.

Testing
Added some more tests.

Squash or Rebase?
Rebase.

Checklist

• I self-reviewed and fully understand this PR.
• WebGPU implementations built with wgpu may be affected behaviorally.
• Validation and feature gates are in place to confine behavioral changes.
• Tests demonstrate the validation and altered logic works.
• CHANGELOG.md entries for the user-facing effects of this change are present.
• The PR is minimal, and doesn't make sense to land as multiple PRs.
• Commits are logically scoped and individually reviewable.
• The PR description has enough context to understand the motivation and solution implemented.

[andyleiserson]

Nice!

There is another instance flag ALLOW_UNDERLYING_NONCOMPLIANT_ADAPTER, which says it "mainly applies to a Vulkan driver's compliance version" (and that is in fact the only place it applies). It might be worth updating the docs for that flag to clarify the difference from STRICT_WEBGPU_COMPLIANCE.

It might be worth a few unit tests (maybe for adapter_allowed) capturing conditions where we don't want to return an adapter. We won't otherwise have much coverage of this logic. An expectation that adapter_allowed exists as a function, is not unused, and has certain properties, provides some protection against breakage.

[andyleiserson]

All alignment-class limits must be powers of 2 is a WebGPU requirement, but this section of the function is applied whether or not the user has asked for strict WebGPU compliance. It would be utter insanity for these not to be a power of two, so it seems fine to reject such devices, but it might be clearer to say something like "Even if the application has not requested strict WebGPU compliance, non-power-of-two alignment limits are nonsensical, so don't attempt to use such a device."

(Side nit: there is a typo in a comment at

wgpu/wgpu-hal/src/dx12/adapter.rs

Line 921 in 523a1cf

// 254

saying the value is 254 rather than 256.)