feat: implement pgvector semantic and hybrid rrf search with bgd indexing

[Prakhar-Sethi012]

Title: feat: Implement Hybrid Search Engine & Background Indexer

Description

This PR introduces the core AI search domain for the WhereTF backend. It sets up the pgvector database infrastructure, implements a background asynchronous indexing service to prevent API blocking, and exposes a high-performance Reciprocal Rank Fusion (RRF) search route that combines semantic vector search with full-text keyword matching.

Type of Change

• New feature (non-breaking change which adds functionality)
• Infrastructure / DB Migration
• Bug fix (Fixed Docker compose command targets and Alembic table naming mismatches)

Architecture & Technical Highlights

• Database Layer: Configured PostgreSQL with the pgvector extension. Created SQLAlchemy models (files and file_content) and generated Alembic migrations to handle vector storage.
• Background Processing: Implemented app/services/indexer.py using FastAPI BackgroundTasks. It opens an isolated SQLAlchemy session, chunks incoming text, generates 384-dimensional embeddings, stores them in PostgreSQL, and cleans up the server's /tmp/ directory.
• Search Engine: Integrated the processing/search.py module to execute complex raw SQL via SQLAlchemy's db.execute().
• Routing: Added the POST /search/ endpoint supporting vector, keyword, and hybrid modes.

How This Was Tested

• Verified local Docker container orchestration (backend safely waits for db health check).
• Resolved module import paths to ensure clean Uvicorn startup.
• Triggered a temporary /test-indexer/ route to ingest a dummy syllabus file.
• Successfully executed POST /search/ (Hybrid Mode) via Swagger UI, receiving a 200 OK with correctly fused vector/keyword scoring and text chunks.

Next Steps & Blockers

• Blocker: The indexer currently relies on a temporary test route.
• Next Action: Awaiting the POST /files/ upload route to be completed so we can route real user files from the web client directly into this background indexing pipeline.