Validate and document the PR review loop end-to-end

[ProtocolWarden]

Context

The two-phase PR review loop (self-review via kodo → human escalation via 👍 or comment) is implemented but hasn't been exercised end-to-end in a documented way. Trust in the write path requires a validation pass and an operator guide.

Goal

Run the PR review loop against at least one real PR (or a controlled test PR) and document the audit trail. Create docs/operator/pr_review.md covering:

• How the self-review phase works (what kodo reads, what LGTM/CONCERNS means)
• How human escalation is triggered
• Guardrail checklist (bot safety contract, <!-- controlplane:bot --> marker, allowed_reviewer_logins)
• What to do when a review gets stuck

Acceptance Criteria

• docs/operator/pr_review.md exists
• At least one end-to-end review run is documented (even if synthetic)
• Guardrail checklist is complete and accurate

Labels

pr-review, operator-docs, trust