drt: dr,kv: drop and restore of a large table

[nicktrav]

Product Area(s): KV, DR.

As seen in https://github.com/cockroachlabs/support/issues/3261, it would be good to exercise the following:

• Simulation of an "accidental drop" of a large table (in the example, operator error resulted in the drop of a ~120TiB table). Ideally this happens in a large, multi-store cluster.
• Restore of the table from a backup.

Having such an operation will allow us to further exercise restore. It will also give us a sense of "time-to-recovery".

This operation will also likely (initially) encounter similar issues to those encountered during scale testing, if run on large multi-store clusters. Notably #134960. The operator will therefore provide a nice before / after view of the improvements we plan to make in this area.

Jira issue: CRDB-49433