[WebGPU] Generate component inventories from fetched Dawn dependencies

[Yang Gu (gyagp)]

Description

Generate a WebGPU component inventory from the Dawn dependencies fetched by a build, then scan and retain it in the Windows, Linux, and macOS plugin packaging pipelines.

• Read the pinned fetcher's dependency list and DEPS without executing them; verify each selected Git checkout's HEAD against its declared commit.
• Record Dawn's archive URL/checksum and, for Windows packaging, the independently distributed DXC archive metadata.
• Run generator tests, write metadata/cgmanifest.json, and scan that directory explicitly because the normal scan excludes the build directory containing fetched dependencies.
• Keep the reviewed manifest and its runtime/development classifications separate. The generated inventory describes build inputs, not which components are linked into the final package.

Motivation and Context

Fetched third-party sources should have a reproducible inventory tied to the actual build. This work is separated from the Dawn upgrade command in #33396 so the two workflows can be reviewed and merged independently.

This PR does not change the Dawn pin, patch stack, runtime code, or reviewed dependency classifications. It has no dependency on the upgrade-tool PR.

Validation

• 12 generator tests passed on an exact snapshot of this branch, including real temporary Git checkouts, missing/stale dependencies, nested metadata, deduplication, and independent DXC metadata.
• Ruff lint/format and Git whitespace checks passed.
• Generator, tests, and pipeline calls are preserved from the combined draft; documentation now describes this workflow independently.

Draft for review of the inventory/scanning contract. The packaging pipelines and Component Governance task have not been executed locally.

[azure-pipelines]

Azure Pipelines:
There may be pipelines that require an authorized user to comment /azp run to run.