Recovery depends on what must survive and which failure the design must handle. Persistent destinations and retained recovery points address different needs from additional compute instances. A regional or zonal label alone doesn't prove that a service can tolerate the stated outage.
The original scenarios ask you to distinguish temporary data, replicas, retained restores and project-level dependencies. Choose the evidence that addresses the actual failure. These decisions don't execute a migration or restore a real disk. A production RTO or RPO claim needs an appropriate plan and a measured recovery exercise, not only an objective written in a document.
Choose protection and evidence for the specific failure.