build(deps): AUT-1608 speed up uv lock resolution

[svcnemo-autobot]

Background

uv lock executes the git-sourced fast-hadamard-transform setup metadata path, which imports Torch/CUDA build helpers and makes resolution unexpectedly slow.

What changed

• Provide static uv metadata for fast-hadamard-transform 1.1.0.
• Record the metadata in uv.lock.
• Guard the metadata contract in the dependency-layer test.

Details

flowchart LR
  A[uv resolver] --> B[static dependency metadata]
  B --> C[skip setup.py execution]
  C --> D[fast lock resolution]

Loading

The declared dependencies match the source setup and existing lock entry: torch, packaging, and ninja.

Tested

• uv lock --check
• bash .github/scripts/test_docker_dependency_layers.sh
• uv run --no-sync pre-commit run --all-files
• Cold resolver: 2m03s before lock update; regenerated resolution: 2.18s; final locked check: 2ms.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[svcnvidia-nemo-ci]

/ok to test 9c42324

[svcnemo-autobot]

All change-relevant checks passed on 9c423243473d3c10f82af09bf8b3c9f97c50d314, including install, lint, copyright, CICD NeMo (unit/H100/GB200), build/release, DCO, secrets, and link checks.

The sole red check is Claude Review pre-flight job 96253092826. It fails at Check NVIDIA SSO membership; the fork-triggered workflow reports Secret source: None, while .github/workflows/claude-review.yml documents that fork pull_request workflows cannot access the secrets/OIDC token required by the review action. An unchanged rerun would therefore be deterministic rather than a flaky retry. Please handle or waive this unrelated workflow check.

[ko3n1g]

@svcnemo-autobot fetch and merge upstream main

[ko3n1g]

@svcnemo-autobot fetch and merge upstream main

[svcnemo-autobot]

💬 The authoring workstation is starting; I’ll merge the current upstream main into the owned PR branch once it is ready.