An operator accidentally deletes rows from RDS for PostgreSQL at 14:10. Automated backups are enabled with a seven-day retention period, and the desired 14:09 recovery point is restorable. The team must recover that earlier state without modifying the current instance until recovery is validated. Which approach should it use?

1. Promote a read replica after the deletion has already replicated to it.
2. Perform an RDS point-in-time restore to 14:09 into a new DB instance, validate it, and plan the required data recovery or application cutover.
3. Take a manual snapshot at 14:20 and restore that snapshot to recover the rows as they existed at 14:09.
4. Fail over to the Multi-AZ standby to obtain the data from before the deletion.