feat(tests): overhaul staging test infrastructure and update fixtures

[tychtjan]

Summary

Added staging test support (PYTHON_DEX), which required rewrites in the test
infrastructure (dynamic PostgreSQL URL replacement instead of duplicating fixtures)
and a new CI workflow for pre-merge staging tests.

Changes by commit

1. chore: regenerate API client from updated OpenAPI spec

• Updated OpenAPI-generated API client with new models and endpoints
• Reviewer: auto-generated, safe to skip

2. feat(ci): add staging test infrastructure and harden test setup

• Add staging-tests.yaml workflow (triggered by label, /test-staging comment, or manual dispatch)
• Overhaul conftest.py for dynamic data source URL replacement (staging vs local)
• Add clean_staging.py and upload_demo_layout.py helper scripts
• Unify token passing: all staging make targets use TOKEN= argument
• Fix LLM provider patch to use InAttributes after spec removed PatchAttributes
• Add alternativeDataSourceId to data source fixture
• Document staging test workflow in CONTRIBUTING.md

3. chore: regenerate test cassettes

• Regenerated all cassettes against staging environment
• Reviewer: regenerated fixtures, safe to skim

Test plan

• Passes local tests
• Passes staging environment tests

JIRA: DX-431
risk: low

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.53%. Comparing base (46c0f4b) to head (038dfb9).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1392      +/-   ##
==========================================
- Coverage   77.53%   77.53%   -0.01%     
==========================================
  Files         225      225              
  Lines       14614    14613       -1     
==========================================
- Hits        11331    11330       -1     
  Misses       3283     3283              

☔ View full report in Codecov by Sentry.
📢 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.

[hkad98]

Suggested change

	- name: Clean staging environment
	run: make clean-staging TOKEN=${{ secrets.PYTHON_SDK_STG_API_KEY }}
	- name: Clean staging environment
	run: make clean-staging
	env:
	TOKEN: ${{ secrets.PYTHON_SDK_STG_API_KEY }}

Either use secrets in env or with.

[tychtjan]

I analyzed why there were some of the bigger deletes in the fixtures

 # OLD: cleanup + verification assertion
  finally:
      sdk.catalog_organization.delete_jwk("demoJwk")
      assert len(sdk.catalog_organization.list_jwks()) == 0  # <-- this GET is now gone

The new pattern uses safe_delete() which only calls the delete, no verification list:

  # NEW: cleanup only, no verification
  finally:
      safe_delete(sdk.catalog_organization.delete_jwk, "demoJwk")

This removes the final GET ...?page=0&size=500 call that verified the entity was gone.
That's why every cassette has a trailing interaction block removed — it was the VCR recording of
that verification GET. This is entirely expected and correct. The verification was fragile on staging (where other entities may exist, making len(...) == 0 fail), and the safe_delete + snapshot/restore approach is more robust.

[hkad98]

This is too broad. Consider checking out .github/workflows/rw-python-tests.yaml and unit-tests.

[tychtjan]

Fixed to run: uv sync --group test --locked