fix: add pg8000 graceful shutdown on Cloud Run scale-down

[panish16]

Summary

• Adds cloud-sql-connector from sbc-connect-common as a dependency in ppr-api and mhr-api
• Calls setup_pg8000_close_event_listener(engine) on app startup to suppress pg8000.exceptions.InterfaceError: network error logged when Cloud Run terminates idle instances and SQLAlchemy's connection pool teardown tries to close already-severed sockets

Services fixed

• ppr-api/src/ppr_api/__init__.py
• mhr-api/src/mhr_api/__init__.py

Background

GCP logs confirmed pg8000.exceptions.InterfaceError: network error occurring in both ppr-api-prod and mhr-api-prod on Cloud Run scale-down. This is the same fix applied across bcgov/lear (PR #4470), bcgov/bcros-common (PR #270), bcgov/sbc-auth (PR #3790), and bcgov/STRR (PR #1669).

Test plan

• Confirm no new pg8000.exceptions.InterfaceError entries in GCP Log Explorer for eogruh-prod after deployment
• Existing unit and integration test suites pass