fix: make committee analysis test assertions less fragile to wording changes

[Copilot]

Tests in test/unit/deep-analysis.test.js hard-coded full English sentences from localized copy, making them brittle to any wording change in COMMITTEE_ANALYSIS_CONTENT_STRINGS.

Changes

• buildCommitteeAnalysis — productivity test: Split '0% active rate signals low legislative productivity' into two independent partial checks so a change to either the template or the productivityLow value only breaks one assertion:

  // before
  expect(result.why).toContain('0% active rate signals low legislative productivity');
  
  // after
  expect(result.why).toContain('0% active rate');
  expect(result.why).toContain('low legislative productivity');
  expect(result.why).not.toContain('moderate');

• buildCommitteeAnalysis — impactPoliticalNone test: Replaced exact-phrase check with a short partial string + baseline comparison, decoupling the test from sentence structure:

  // before
  expect(result.impactAssessment.political).toContain('No committees have published recent documents');
  expect(result.impactAssessment.political).not.toContain('Active committees');
  
  // after
  const baseline = buildCommitteeAnalysis(COMMITTEE_DATA, '2026-02-24');
  expect(result.impactAssessment.political).toContain('No committees');
  expect(result.impactAssessment.political).not.toBe(baseline.impactAssessment.political);

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.