Usability improvement on manipulating pd-ctl manually

Feature Request

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

Currently, if we want to do certain operations using pd-ctl, typically we need to log into a pd pod, figure out the certificates and host information, and then file the command. There're two problems with this approach:

1. It could be hard to automate
2. It could be difficult or impossible in some restricted environment where users are not given permissions to exec in pods

Describe the feature you'd like:

It could be helpful if we have some resources to automate the PD operations, maybe something similar to TidbInitializer, which helps us on automating the TiDB initialization by running SQL statements for us. By exposing a CRD for PD operations, it would be helpful for unifying the privilege management to k8s's RBAC, and it would be easier to perform certain automations or gitops operations.

Describe alternatives you've considered:

The alternative is just logging into PD pods and manually run commands using pd-ctl. We need to figure out all the connection info (certs, keys, hostnames, etc) on our own and it's could be hard to automate.

Teachability, Documentation, Adoption, Migration Strategy:

Could be similar to TidbInitializer, could use that as a reference.

Start by reviewing how pd-ctl currently obtains PD connection information and use TidbInitializer as the reference for the proposed resource. Define the PD operations and CRD behavior that need to be supported, including Kubernetes RBAC and restricted environments. Done means users can automate the requested PD operations without exec access to PD pods.