Adding Postgres Support

[AbeOmor]

This will address: #136

This pull request introduces PostgreSQL support with pgvector integration for vector similarity search, replacing the default retrieval endpoint and adding new configurations, scripts, and dependencies.

Key changes include updating configuration files, adding PostgreSQL-specific scripts, and modifying the retrieval logic to support the new database type.

PostgreSQL Integration

• Added a new PostgreSQL connection string in .env.template for database configuration.
• Updated config_retrieval.yaml to set PostgreSQL as the preferred retrieval endpoint and added configuration details for the postgres endpoint.
• Introduced setup_postgres.sql and setup_postgres_schema.py scripts for setting up and validating the PostgreSQL schema with pgvector.
• Added a new example script, postgres_example.py, to demonstrate PostgreSQL vector database integration.

Retrieval Logic Updates

• Modified retriever.py to include a new client, PgVectorClient, for handling PostgreSQL-based retrieval.
• Added a diagnostic script, postgres_diagnostics.py, to identify and resolve PostgreSQL configuration issues.
• Added a test connect script, test_postgres_connection.py, to identify and resolve PostgreSQL configuration issues.

Dependency Updates

• Added psycopg2-binary and pgvector to requirements.txt for PostgreSQL integration and vector operations.

Future Opportunities:

• Update to psycopg3 (Psycopg3 support AbeOmor/NLWeb#1)

Working demo.

https://www.loom.com/share/b7c14ff1f0b347a38f51b15ea5175701?sid=31410d68-1bb8-41aa-9e24-dc6f951396ca

[AbeOmor]

@chelseacarter29 and @rvguha Just publish Postgres support PR, it is out of draft