[docdb] Batch yb_backup.py's yb-admin commands to improve backup performance

[hulien22]

Jira Link: DB-1151
We currently perform this step during a restore (from issue #8892):

 12 min: PHASE 2: Generate list of TSes for every tablet
 19 min: PHASE 3: Find all table/tablet data dirs on all TSes

Phase 2 currently involves sending a yb-admin list_tablet_servers for each tablet individually (29d2c2c added some parallelization to this, but this can be further improved by batching all these requests into a singular yb-admin request).

Additionally, we currently need to perform a second request to all tservers for each table to find their tablet directories for Phase 3.

Ideally, we should combine these two phases to one, with a single batched yb-admin command that can retrieve the tablet servers for the list of tablets, and also return the list of corresponding data directories. This requires a new endpoint for yb-admin, and adding new parsing support for yb_backup.py.

[hulien22]

One other thing to consider is that the backup script needs to determine if the db has this capability, and if so decide to use this new yb-admin command, else use the old approach