Detect blocked Master replication threads and dump diagnostic logs pre-switchover

During a replication lag event where the Master node falls behind its Slaves, automated failover tools (like MaxScale) trigger a switchover before administrators can collect transient diagnostic data. This feature should provide automated, proactive detection and logging on the Master before the switchover occurs to ensure root-cause visibility.
Requirements

a) Blocked Thread Detection: Implement a mechanism to actively detect when a replication thread on the Master node has stalled or become blocked.

b) Pre-Switchover Log Dump: Immediately upon detection—and prior to a cluster switchover—automatically dump all required diagnostics (e.g., thread stack traces, InnoDB status, and process lists) to find the exact cause of the hang.

This specific replication-stall scenario should be incorporated as part of the following existing enhancement requests:
Ticket 223068 / Jira MDEV-38123 - Enhancement request to dump required logs when server crashes
Ticket 236942 / Jira MENT-2637 - MariaDB cluster health audit tool