fix(gym): make the gym venv prefetch dry run pass again

[yfw]

Summary

• Backport 48d71a3 (fix(gym): make the gym venv prefetch dry run pass again #3785) from main onto the current Super v3.5 branch after its Gym bump.
• Supply required Gym VLLM model fields in the Super and Ultra prefetch configs.
• Use hardlinks inside the Gym actor so setup scripts using uv --no-cache remain compatible, and add regression validation against Gym config models.

Validation

• Gym config tests: 7 passed, 2 skipped because the optional vllm extra is unavailable, as documented by the test
• Targeted pre-commit hooks: passed, including Ruff, Ruff format, Pyrefly, whitespace, and recipe minimization checks
• git diff --check: passed

Known limitations

• The full Docker image prefetch was not run locally; the committed test documents the remaining local-vLLM and UV link-mode coverage gaps.

[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.

[yfw]

/ok to test d5ab670