[server] Add per-task timeout for RebalanceManager to prevent indefinite blocking

[swuferhong]

Purpose

Linked issue: close #3096

• Fix a liveness issue where RebalanceManager could block indefinitely if a rebalance bucket's ISR never converges (e.g., target replica is on a dead TabletServer).
• Introduce a per-task timeout (2 minutes) with periodic checks (every 30 seconds). When an in-flight bucket exceeds the timeout, it is treated as COMPLETED, unblocking subsequent buckets and the overall rebalance.
• Add RebalanceTaskTimeoutEvent to the coordinator event system for safe single-threaded state mutation.

Brief change log

Tests

API and Format

Documentation

[loserwang1024]

I left a comment

[LiebingYu]

@swuferhong Thanks for the PR. I have some comments.