Optimize PostgreSQL connection handling (introduce pooling or shared connection)

[snabo1988]

Problem

The current implementation creates a new PostgreSQL connection for each query
via psycopg2.connect().

While acceptable for early development, this may become inefficient under load
and should be optimized.

Current State

• get_connection() opens a new DB connection on every call
• No connection reuse or pooling is used

Proposed Improvements

• Introduce connection pooling (e.g. psycopg2.pool or similar)
• Or maintain a shared connection managed by the application lifecycle

Notes

This is not critical for MVP correctness but should be addressed before
production or higher load scenarios.