PD & TiDB only removes the failure members until all of the failed pods recover

Feature Request

Is your feature request related to a problem? Please describe:

PD & TiDB only removes the failure members in tc status until all of the failed pods recover, however, it's possible that only part of the failed pods recover, in which case, we can remove the failure members for the recover pods instead of waiting for all of the pods to recover.
We need to improve the failover logic.
Describe the feature you'd like:

Describe alternatives you've considered:

Teachability, Documentation, Adoption, Migration Strategy:

Start by tracing the PD and TiDB failover logic that removes failed members and examine how it handles pod recovery. Reproduce a case where only some failed pods recover; done means the corresponding recovered members are removed without waiting for every failed pod to recover.