fix: configure SQLAlchemy connection pool with env-driven settings

[snowrugar-beep]

Summary

Configures SQLAlchemy engine pool settings (pool_size, max_overflow, pool_recycle, with pool_pre_ping always enabled) on both engine construction sites in the project. Eliminates the silent defaults that previously allowed connection exhaustion and stale-connection failures under load.

Changes

• New helper _engine_pool_kwargs() in web_app/db/database.py (single source of truth).
• New init_engine(db_url) factory; both init_db and DBConnector.__init__ now route through web_app.db.database.create_engine with the shared pool policy.
• New tests/test_db_pool.py verifying both engine sites forward the env-driven kwargs and fall back to defaults.
• Dropped the now-redundant import os / import create_engine from web_app/db/crud/base.py.
• Defaults: DB_POOL_SIZE=5, DB_MAX_OVERFLOW=10, DB_POOL_RECYCLE=1800. All overridable via env.

Testing

• python -m py_compile clean across database.py, crud/base.py, and the new test file.
• pytest web_app/tests/test_db_pool.py validates kwarg forwarding under both set-env and unset-env scenarios.
• DRY: a future change to defaults happens in exactly one place.

Closes #59