feat: implement issue #2495 — pr-review: Fable adjudication tier before escalating to a human
Problem
pr-review: Fable adjudication tier before escalating to a human
From the issue: The pr-review cascade hands a PR to a human (needs-human-review + CODEOWNERS request) whenever cascade-action returns decision=escalate (scripts/post-pr-review.sh, the DECISION = escalate branch) or the review-cycle cap is reached (scripts/review-one-pr.sh, "human attention needed"). Both are model-judgment outcomes, and they reach a human with no second opinion from a stronger model. Over the last 30 days that was roughly 1–2 escalations a day across the org.
Risk
Medium — changes GitHub Actions workflow behavior, which is exercised only post-merge; verify via the affected workflow runs.
Test plan
Tests added/updated: tests/test_adjudicate.bats,tests/test_model_sandbox.bats tests/test_post_pr_review_escalation.bats,tests/test_review_one_pr_adjudicate.bats. Verification: bash scripts/dev-lead-lint.sh (shellcheck --severity=warning) ran pre-commit; the bats suite runs in CI.
Rollback
Revert this PR. No non-revertible side effects (no tags, migrations, or external state).
Monitoring
Watch the affected workflow run(s) in the Actions tab and this PR's Lint check for regressions.
Closes #2495