[E-R-Butch]

fix(hindsight): warn when daemon fails due to stale pg0 PID

Summary

Resolves #110838.

When the Hindsight embedded PostgreSQL daemon fails to start after an OS reboot due to a stale PID (vectorize-io/pg0#37), _daemon_start_worker() in plugins/memory/hindsight/__init__.py previously caught the exception and silently logged it only into $HERMES_HOME/logs/hindsight-embed.log. As highlighted in #110838, _check_local_runtime() still reported the runtime as available, leaving users unaware of the silent failure.

Following the discussion with @KeyArgo in #110838:

• Classify Failed to start embedded PostgreSQL and Instance already running errors in _daemon_start_worker().
• Emit an actionable logger.warning and a console stderr notice (⚠ ...) directing users to verify port 5432 and inspect postmaster.pid.
• Add a dedicated unit test in tests/plugins/memory/test_hindsight_provider.py.

[alt-glitch]

This was generated by AI during triage.

Duplicate of #110845 (earlier open PR, same mechanism: classify the pg0 Instance already running failure in _daemon_start_worker and surface a warning with recovery guidance for #110838). Maintainer picks the implementation.

[E-R-Butch]

Closing: fixed at the source — vectorize-io/pg0#38 (pg0 0.15.2) handles pid reuse after reboot, making this warning layer moot. See #110838.