Add Proofline to agent-specific evaluation

[ceodaradigu]

What changed

Adds one annotated Proofline entry to §9, Agent-specific evaluation.

Why it belongs

Proofline evaluates whether agent work is actually supported by explicit requirements and fresh evidence. Its deterministic gate produces NEEDS_EVIDENCE, CONFLICT, APPROVAL_REQUIRED, or READY, then emits a hash-addressed proof packet. The repository includes runnable synthetic fixtures and a public Google ADK / Cloud Run demo.

This is deliberately not presented as a model-quality evaluator. Its scope is task-completion evidence and the approval boundary before an external action.

Disclosure and maturity

This is a self-submission from the RELAUNCH DEPT maintainer. The project is new; the entry explicitly states that there is no independent adoption claim. This PR was prepared with OpenAI Codex assistance under RELAUNCH DEPT's accountable review workflow.

Verification

• Checked the README, repository code search, and all-state pull requests for Proofline and the canonical repository URL; no duplicate was found.
• Verified the canonical repository is public, active, unarchived, and MIT-licensed.
• Read CONTRIBUTING.md and matched its one-line why, URL, section, evidence, and caveat requirements.
• git diff --check passes.
• Changed only README.md.