OpenShift: Migrate prod MinIO to HA distributed topology

## Summary
Prod MinIO currently runs as a single-replica deployment backed by one RWO PVC. This limits availability and prevents horizontal scaling.

## Scope
- Replace single-instance MinIO deployment with distributed MinIO (StatefulSet) using per-pod PVCs
- Keep a single stable service/route endpoint for clients
- Validate object PUT/GET behavior during and after cutover
- Plan and execute data migration from existing bucket data to distributed cluster
- Define rollback procedure

## Acceptance Criteria
- MinIO runs with 4+ pods and per-pod storage
- Client endpoint remains stable
- Existing objects are accessible after migration
- Runbook documents cutover + rollback

## Impact
Reduces downtime risk and removes single-pod storage bottleneck in prod.