chore(metadata): regenerate metadata.json, skills.sh.json, and benchmarks.json

[github-actions]

Auto-regenerated by generate-skill-metadata workflow.

• Trigger: workflow_run
• Generator log: see workflow run.

cc @jim-nvidia @jasonnvidia

[mosheabr]

Superseded by #400.

This PR was generated by the pre-fix aggregator, so it records the v2 reports as has_results: false with null tasks, no agents and no uplift. #400 fixes the parser to read both the v1 and v2 BENCHMARK.md layouts and ships a correctly regenerated benchmarks.json, which is now on main.

Merging this would make the blanked entries authoritative, so closing rather than rebasing. The metadata workflow will open a fresh PR on its next run if any drift remains.