fix: prevent leader pod from being recreated twice on background deletion (#805)

When a leader pod with RestartPolicy RecreateGroupOnPodRestart is
deleted using the default background propagation, stale worker pods
from the old group can trigger an unintended second leader recreation.

Add workerPodBelongsToLeader() to verify the ownership chain of a
worker pod back to its current leader before allowing it to trigger
group recreation. This prevents background-deleted workers from
causing duplicate leader pod creation.

Signed-off-by: Kay Yan <kay.yan@daocloud.io>