[DocDB] Newly RBSed tablet replica spurious delete race #32434

Jira Link: DB-22101

Description

Caught by TabletReplacementITest.DontDeleteNewReplicaInPendingConfigAfterRbsFromFollowerRf5 test in TSAN build on top of 2024.2.
Reproducible even before commit adding that test. Probability is very low because need to met all of the following conditions:

1. Master leader sent DeleteTablet RPC to tserver before RBS is completed - this can happen but usually harmless because tserver-side has its own guard A to reject it if tablet replica is in RBS process.
2. DeleteTablet RPC reached tserver after RBS is competed but before tserver->master heartbeat with Raft config pending op id is received by master leader. Then guard A is not longer protecting tablet replica. And master-side guard B is not yet effective.

Issue Type

kind/bug

Warning: Please confirm that this issue does not contain any sensitive information

• I confirm this issue does not contain any sensitive information.