Remove the AI analysis feature

[MaxGhenis]

Closes #3800.

Removes the AI analysis feature from API v1 rather than patching it. Nothing live calls it — policyengine-app-v2 has no references to /simulation-analysis, /tracer-analysis or /ai-prompts (only the archived v1 app did) — and the model it pinned (claude-sonnet-4-20250514) is retired (Models API returns 404). The feature will be rebuilt properly during the Axiom migration.

Removed

• Routes: simulation_analysis_routes, tracer_analysis_routes, ai_prompt_routes and their registrations in api.py; the matching OpenAPI paths/schemas and migration-contract entries (migration_registry.py, docs/generated/migration_contracts.json).
• Services: ai_analysis_service, ai_prompt_service, simulation_analysis_service, tracer_analysis_service; the ai_prompts/ and ai_templates/ packages; the ai-analysis runtime-cache family; the tracer-analysis payload validator.
• The anthropic dependency (pyproject + lock) and every ANTHROPIC_API_KEY deployment hook: Cloud Run env/secret mapping, secret sync, App Engine app.yaml/export, the deploy-env validators, and the PR/Push workflow env. The GCP secret itself is untouched (manual cleanup whenever).
• The tests, fixtures and snapshots that only covered the deleted code (including the SDK contract test from Pin anthropic below 1.0 and guard the SDK call contract #3799, which has nothing left to guard).

66 files, +26 / −2,840 (includes removing services/run_sync_utils.py, orphaned since May's Stage 7 refactors — no importers, no tests).

Deliberately kept (separate follow-ups)

• The analysis table and its Analysis ORM model — dropping it is a schema change and the repo's rule is autogenerated Alembic revisions against a live MySQL, so it gets its own migration PR.
• /calculate's household-trace storage (runtime_cache/household_traces.py, household_calculation_service.py) — its only consumer was tracer-analysis, so it's now dead weight per calculation, but it touches the most important endpoint and deserves its own review.

Verification

• CI: all checks green; codecov project 85.16% → 85.48% (+0.32%), patch 100%.
• tests/unit: 1,040 passed (only the pre-existing local-only test_gcp_logging.py collection artifact excluded); tests/contract: 36 passed; migration-artifact tests pass; import policyengine_api.asgi_factory, policyengine_api.migration_registry ok; ruff format/check clean.
• git grep -i anthropic now hits only CHANGELOG.md, docs/migration/history/ and the changelog fragment.
• Built by Sol from a written spec; reviewed and lock regenerated by Claude. Draft pending the maintainer's decision — merging deploys production.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.48%. Comparing base (a2880ef) to head (e117906).
⚠️ Report is 6 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3801      +/-   ##
==========================================
+ Coverage   85.16%   85.48%   +0.32%     
==========================================
  Files         110       95      -15     
  Lines        6309     5897     -412     
  Branches     1096     1030      -66     
==========================================
- Hits         5373     5041     -332     
+ Misses        630      559      -71     
+ Partials      306      297       -9     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.