Streaming Endpoints for VA-Spec Annotations

[bencap]

This pull request refactors the endpoints in score_sets.py that return annotated variant data for a score set. Instead of returning the entire result as a single JSON object, these endpoints now stream results as newline-delimited JSON (NDJSON), improving performance and scalability for large datasets. The implementation introduces a reusable streaming generator and updates endpoint documentation, response headers, and eager loading of related data.

Streaming and API Response Refactor:

• Introduced _stream_generated_annotations, a generator function that streams annotation results as NDJSON, including progress logging and summary information. This function is now used by all three annotation endpoints. (F51675eeR479)
• Updated the three annotation endpoints (/annotated-variants/pathogenicity-evidence-line, /functional-impact-statement, /functional-study-result) to return a StreamingResponse with NDJSON instead of a single JSON object, and added appropriate response headers for metadata and CORS. (F51675eeR568, F51675eeR681, F51675eeR792, F51675eeR874)

Query and Performance Improvements:

• Refactored database queries for mapped variants in all three endpoints to use .join and .options with contains_eager and selectinload for more efficient eager loading of related data (e.g., publications, users, licenses, experiments). (F51675eeR624, F51675eeR735, F51675eeR850)

Documentation and Typing:

• Expanded endpoint docstrings to clearly describe the new streaming NDJSON response format, headers, and error handling. (F51675eeR489, F51675eeR603)
• Added TypedDict response types for each annotation stream to clarify response structure. (F51675eeR479, F51675eeR648, F51675eeR759)

Code Clean-up and Import Sorting:

• Improved import organization and sorted imports for clarity and consistency. [1] [2] [3] [4]
• Minor update to test utility imports for consistency.