Expand unit test coverage for metric-engine and rule-engine

[Copilot]

The core evaluation engines (metric-engine.ts, rule-engine.ts) were only exercised indirectly through antipatterns-e2e.test.ts, leaving regressions invisible to the test suite.

Changes

• metric-engine.test.ts (+20 tests)

  • evaluateMetric edge cases: empty datasets, no matches, missing/null fields, default example fallback, avg(field) aggregation, 3-row example cap.
  • Threshold-override coverage for evaluateTrigger, calibrate, and runTestCases.
  • calibrate edge cases: empty rows, non-numeric filter field (null distribution), sessions scope derivation.
  • serializeCalibration / stripCalibration round-trip, including the no-distribution and no-op branches.
  • parseRuleExtensions with inline-array requires, malformed JSON test cases (skipped silently), and bodies without frontmatter.
  • parseMetric defaults (scope, aggregation, version) and missing-id rejection.

• rule-engine.test.ts (+1 test)

  • evaluateRule returns pct: '0%' when count > 0 but total === 0 (avoids divide-by-zero in templates).

Coverage on metric-engine.ts moves from ~86% lines to near-complete; rule-engine.ts remains at ~98% with the divide-by-zero branch now pinned.

[github-actions]

⚠️ Deprecation Warning: The deny-licenses option is deprecated for possible removal in the next major release. For more information, see issue 997.

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Scanned Files

None