fix: fallback to replset primary when leader lease expired

[wallyxjh]

• Symptom: Scaling MongoDB from 3 replicas down to 1 gets stuck; Component controller keeps logging cluster has no leader. Cluster spec shows replicas: 1 while InstanceSet remains at 3.
• Root cause: MongoDB does not run the HA loop by default, so the leader ConfigMap lease expires. leaveMember relies on the leader CM to locate the primary and fails when the leader is missing.
• Fix: Add a fallback in the MongoDB lorry manager: when the leader CM is invalid, discover the primary via replSetGetStatus; if needed, probe members directly to find the primary, then continue
  replSetReconfig.
• Impact: Limited to MongoDB scale‑in/leaveMember behavior (lorry sidecar). No behavior change when leader CM is valid.
• Relevant code: pkg/lorry/engines/mongodb/manager.go
• Verification:
  1. Scale out to 3: kubectl patch cluster -n --type json -p '[{"op":"replace","path":"/spec/componentSpecs/0/replicas","value":3}]'
  2. Expire leader lease: kubectl patch cm -n -mongodb-leader --type merge -p '{"metadata":{"annotations":{"renew-time":"1","acquire-time":"1","ttl":"15"}}}'
  3. Scale in to 1: patch replicas back to 1
  4. Old image gets stuck with cluster has no leader; new lorry image completes scale‑in and logs are clean.