feat: Add agent skills for NeMo Gym

[lbliii]

Summary

• Adds 7 agent skills following the agentskills.io spec: gym-review, gym-debug, gym-profile, gym-config, gym-data, gym-scaffold-agent, and updates to add-benchmark
• Each skill includes evals/evals.json with 3 assertion-based evals (21 total) and a chains.yaml for multi-step workflows
• gym-review is an S-tier reference implementation with a standalone deterministic Python checker (scripts/review.py), self-contained anti-pattern/fix-pattern references, and portable eval fixtures — works without the NeMo Gym repo
• Skills cover the full contributor workflow: data prep, config composition, server scaffolding, code review, debugging, and reward profiling

Test plan

• Run python .claude/skills/gym-review/scripts/review.py .claude/skills/gym-review/evals/files/ and verify expected findings per fixture
• Verify skills load correctly in Claude Code (check / menu shows all 7)
• Run with-skill vs without-skill evals per agentskills.io eval spec

🤖 Generated with Claude Code

[lbliii]

Superseded by new PR — original branch couldn't be force-pushed to fix DCO sign-off.