During pod termination (hibernation, rolling update, node drain),
PostgreSQL's built-in archiver may be killed before it can archive
the final WAL segment. Add a call to ArchiveAllReadyWALs() after
each TryShuttingDownSmartFast() in both shutdown paths (context
cancellation and SIGTERM signal) to sweep any remaining .ready WAL
files via the plugin sidecar, which outlives the postgres container
as a native sidecar.

A fresh context.Background() is used for the WAL sweep because the
parent context may already be cancelled in the context-cancellation
path, which would cause gRPC calls to the sidecar to fail
immediately.

Errors are logged but do not prevent the instance manager from
exiting, consistent with existing shutdown error handling.

Closes cloudnative-pg#10341

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Signed-off-by: Andrew Gleeson <Andrew_Gleeson@external.mckinsey.com>