Purpose

Resolves all review comments on #242 (the open dev -> main PR).

CodeQL (github-code-quality) — unused imports / variables

• src/processor/src/tests/unit/services/test_queue_service_internals.py
  • Drop redundant tp = _task_param() in test_output_cleanup_refuses_broad_prefix.
  • Drop unused _no_op_worker in test_start_service_runs_and_completes.
• src/processor/src/tests/unit/libs/base/test_application_base_init.py
  • Drop unused import inspect.
  • Drop unused app local in test_init_loads_app_config_when_url_set.
• src/processor/src/tests/unit/libs/agent_framework/test_agent_speaking_capture.py
  • Drop unused datetime / MagicMock imports.
• src/processor/src/tests/unit/utils/test_agent_telemetry.py
  • Drop unused from datetime import UTC, datetime (only the literal UTC string is used).
• src/processor/src/tests/unit/utils/test_agent_telemetry_records.py
  • Drop unused import pytest.
• src/backend-api/src/tests/sas/storage/blob/test_config.py
  • Drop unused pytest import and unused default_config symbol (still referenced via blob_config_module.default_config).

Copilot review — order-dependent MagicMock leak

• src/backend-api/src/tests/sas/storage/blob/test_async_helper_extra.py
• src/backend-api/src/tests/sas/storage/blob/test_helper_extra.py

In _wire_credential, the tests were mutating type(cred).__name__ on a MagicMock(). Because every MagicMock() instance shares the same MagicMock class, that assignment globally renames MagicMock — so any later test that introspects type(...).__name__ sees whatever the last test set, producing order-dependent / flaky behavior.

Replaced the mutation with a dedicated dynamically-created stub class per credential type (e.g. type('DefaultAzureCredential', (), {})). This:

• Scopes __name__ to a fresh class per call (no leakage).
• Preserves hasattr(cred, 'account_key') semantics for both the account-key path (attribute set on the stub) and the AAD path (attribute absent).
• Keeps all existing call sites unchanged.

Verification

• cd src/backend-api && PYTHONPATH=src/app pytest src/tests -c NUL --rootdir=. --cov=src/app --cov-fail-under=82
  • 585 passing, 93.28% coverage. (1 unrelated env failure when locally authenticated against the placeholder example.azconfig.io; passes in CI where no Azure credentials are present.)
• cd src/processor && pytest src/tests --cov=src --cov-fail-under=82
  • 812 passing, 87.44% coverage.

Does this introduce a breaking change?

• Yes
• No

Golden Path Validation

• Test-only change; backend + processor unit suites continue to pass and meet the 82% coverage gate.

Deployment Validation

• No runtime / deployment code touched.