Consider a better way to get the max-alloc-id when recovering PD cluster

## Feature Request

### Describe your feature request related problem

If the` max-alloc-id` is set incorrectly when recovering PD cluster, it will cause duplicate region id and cause the cluster to be unavailable. We currently use monitor and log to find it, but it is possible that people forget to set or the ID found in the log is incorrect（may log loss). We need a safer way to get the id.

### Describe the feature you'd like

### Describe alternatives you've considered

consider let recover tool do not need to set the id or a tool can more easy and safe to get the id.
- consider scanning all tikvs to get id
- consider collecting all regions' heartbeat then turn on `allocer`
and so on../
### Teachability, Documentation, Adoption, Migration Strategy
- Go