Fix upstream simulation error responses

[anth-volk]

Fixes #2430

Summary

• Return HTTP 502 when the simulation service reports a failed execution.
• Preserve the upstream failure message in the API response.
• Return cached failed executions as errors instead of successful responses with a null result.
• Keep successful and in-progress responses at HTTP 200 with a null message.
• Add a changelog fragment.

Testing

• Added focused service coverage for immediate failures, cached failures, and cached failures without a stored message.
• Added focused route coverage for failed, successful, and in-progress responses.
• 134 focused economy service and route tests passed.
• 38 API v1 route-contract tests passed.
• Ruff formatting and lint checks passed.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 10 lines in your changes missing coverage. Please review.
✅ Project coverage is 46.21%. Comparing base (37e168a) to head (9122964).

Files with missing lines	Patch %	Lines
policyengine_api/services/economy_service.py	0.00%	10 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3821      +/-   ##
==========================================
- Coverage   46.26%   46.21%   -0.05%     
==========================================
  Files         135      135              
  Lines        7963     7971       +8     
  Branches     1326     1328       +2     
==========================================
  Hits         3684     3684              
- Misses       3990     3998       +8     
  Partials      289      289              

☔ 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.

[anth-volk]

Deployment validation for this PR currently depends on resolving the pre-existing failure in tests/integration/test_live_phase10_staging.py::test_live_phase10_activation_failure_and_retry on master (failed workflow run). This PR should remain blocked until the existing deployment sequence completes successfully.