Complete SV-Bench v0.1 planning gates

[intertwine]

Summary

• persisted the attached issue plan under plans/ and implemented the SV-Bench v0.1 / suite-v1 planning gates
• added manifest, reward-comparison, config-validation, beta-environment, and checklist tooling
• documented current status, roadmap, leaderboard schemas, public mini data policy, and E3-E6 beta surfaces
• updated Prime lab readiness handling so hosted auth failures report fallback-ready without blocking local validation

Validation

• uv run ruff check .
• make check
• uv run python scripts/validate_svbench_configs.py
• make svbench-v0.1-check
• make suite-v1-check
• make eval-e3 N=10
• make eval-e4 N=10
• make eval-e5 N=10
• make eval-e6 N=10
• make lab-check (fallback-ready; Prime auth unavailable locally)
• uv lock --check

Secondary review

• incorporated independent review findings from secondary agents Kant and Pauli, including stronger E3-E6 validation, self-contained reward-source configs, stricter public-data scans, and sv_shared packaging fixes.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f6e661970d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[intertwine]

Addressed Codex review feedback: Docker-mode ToolRunner now passes caller-provided env names through with docker --env NAME while avoiding secret values in the recorded command, with regression coverage in scripts/tool_runner_rendering_test.py. Local make check passes after the fix.