Add ClinicalMultimodalDB, Flask clinical API, docs and end-to-end tests

[pifu-dabai]

Motivation

• Provide a lightweight, notebook-friendly multimodal clinical database with standardized schema and retrieval/analysis utilities.
• Expose ingestion, extraction, QA, search and research queries via a simple Flask API for integration and testing.
• Document the data management / research engine plan to guide implementation and extension.

Description

• Add database/clinical_multimodal_db.py implementing ClinicalMultimodalDB and SearchResult with schema, FTS5 semantic search, regex-based auto-fill, RAG-style answer synthesis, IQA scoring, change detection, knowledge-graph extraction, lesion evolution and research query helpers.
• Export ClinicalMultimodalDB and SearchResult from database/__init__.py.
• Add server/clinical_api.py which provides a Flask application via create_app with endpoints for health, upsert patient, visits, medications, lesions, labs, assets, auto-fill (single/batch), IQA, semantic search, RAG, multimodal ingest, review commit, change detection, snapshot, graph extraction, evolution, audits and research queries.
• Add implementation plan documentation in docs/data_management_research_engine_plan.md describing goals, data model, intelligent fill, RAG/graph/IQA features, architecture and phased rollout.
• Add comprehensive integration tests in tests/test_clinical_api.py that exercise end-to-end flows and validation guardrails using a temporary SQLite database and the Flask test client.

Testing

• Ran the test module tests/test_clinical_api.py with unittest using a temporary DB and Flask test client.
• The end-to-end flow test exercises patient/visit ingestion, auto-fill, batch fill, medications/lesions/labs ingestion, semantic search, RAG, IQA scoring, change detection, snapshot, review commit, graph extraction, evolution, audits and research query.
• All automated tests in tests/test_clinical_api.py completed successfully.

---

Codex Task