Add validation tests for mock LLM responses in test_beatrica.py

[chigwell]

User Story
As a software developer,
I want to enhance the test suite in tests/test_beatrica.py with mock LLM response validation
so that we ensure review aggregation logic correctly processes both valid and invalid LLM outputs.

Background
The current test suite uses basic mocks that don’t validate the structure or content of LLM responses. This leaves gaps in verifying whether the aggregation logic in beatrica.py (specifically the aggregate_review_points handling) correctly parses and processes LLM outputs. The test_review_initialization_openai test in tests/test_beatrica.py only checks basic initialization, not response validation. Without structured validation, malformed LLM responses (e.g., missing XML tags, incorrect patterns) could lead to silent failures or incorrect review outputs.

Acceptance Criteria

• Modify tests/test_beatrica.py to include mock LLM responses that simulate:
  • Valid XML-structured reviews (e.g., <aggregated_review>...</aggregated_review>)
  • Invalid responses (missing XML tags, unparsable patterns, empty content)
• Add test cases validating that:
  • The aggregate_review_points logic correctly extracts reviews from valid XML
  • Invalid responses trigger appropriate error handling (e.g., logging, skipped aggregation)
  • Edge cases (e.g., partially valid XML, multiple <aggregated_review> tags) are handled
• Update mocks in test_review_initialization_openai to use realistic LLM response structures from prompts.py
• Ensure all existing tests pass after changes, confirming no regressions in core functionality

Validation Steps

1. Run pytest tests/test_beatrica.py -v and verify new tests pass
2. Check logs for proper error messages when invalid LLM responses are simulated
3. Manually inspect aggregated outputs in test runs to confirm alignment with mock response content