Add engine-neutral execution receipts to API v1

[MaxGhenis]

Summary

• add the engine-neutral execution receipt v1 compatibility serializer and strict validator, with RFC 8785 canonical request/result hashes and certified PolicyEngine bundle identities
• propagate receipts through POST /{country}/calculate and GET /{country}/household/{household_id}/policy/{policy_id} while preserving the existing top-level result shape
• persist household receipts with their results in a versioned JSON cache envelope; legacy cache rows remain readable and cached receipt hashes are verified before replay
• validate economy receipts against both the returned result and available gateway execution context, including run, runtime, model, population, data-build, and resolved-app identities
• preserve valid native Axiom receipts without relabeling them as PolicyEngine, and synthesize PolicyEngine receipts only when concrete runtime and population evidence is available
• document the strict receipt schema in OpenAPI and extend migration contracts with optional stable response fields so receipt omission remains compatible

Compatibility and failure behavior

Execution provenance is additive and fail-open. A completed household or economy result still returns when receipt construction or verification fails, and legacy cached results are never assigned fabricated provenance. The cache envelope uses the existing JSON column, so this change does not require a database migration.

Optional migration-contract fields protect the receipt shape when it is present without making provenance mandatory for workers or historical rows that cannot prove their execution identity.

Axiom and Populace migration

The receipt contract separates requested aliases from resolved engine, model, ruleset, and population artifacts. Economy enrichment preserves valid Axiom-native receipts and removes contradictory PolicyEngine metadata. Population provenance is emitted only for a concrete URI plus data version/build identity, providing a truthful handoff point for Populace-backed workers rather than treating a dataset selector as an artifact identity.

Validation

• full test suite: 924 passed, 11 skipped
• latest focused receipt, household, economy, and migration-contract suite: 206 passed
• Ruff check and format check on changed Python files
• git diff --check

[codecov]

Codecov Report

❌ Patch coverage is 67.07317% with 162 lines in your changes missing coverage. Please review.
✅ Project coverage is 79.12%. Comparing base (21427fa) to head (8541ea6).
⚠️ Report is 6 commits behind head on master.

Files with missing lines	Patch %	Lines
policyengine_api/execution_receipt.py	63.73%	63 Missing and 65 partials ⚠️
policyengine_api/services/economy_service.py	68.75%	23 Missing and 7 partials ⚠️
policyengine_api/constants.py	60.00%	1 Missing and 1 partial ⚠️
policyengine_api/endpoints/household.py	94.73%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3750      +/-   ##
==========================================
- Coverage   79.92%   79.12%   -0.80%     
==========================================
  Files          70       71       +1     
  Lines        4338     4815     +477     
  Branches      808      937     +129     
==========================================
+ Hits         3467     3810     +343     
- Misses        652      709      +57     
- Partials      219      296      +77     

☔ 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.