fix hung in remote lock if cn restart quickly to 1.0-dev

[zhangxu19830126]

What type of PR is this?

• API-change
• BUG
• Improvement
• Documentation
• Feature
• Test and CI
• Code Refactoring

Which issue(s) this PR fixes:

issue #12554

What this PR does / why we need it:

Fix hung in remote lock if cn restart quickly, it will call remote lock on rpc io goroutine, and block all remote requests.

Assuming that we have cn0, cn1, and table1, we consider the following timing:

1. at time t0, cn0 obtains the t1 lock table, and the lock-table bind is t1-cn0-table1-version1.
2. at time t1, cn0 down.
3. at time t2, cn0 restarted, and (t2-t1) < cfg.KeepBindTimeout，so lock-table allocator will keep
   the bind t1-cn0-table1-version1 valid
4. cn1 try to lock table1 and gets the binding t1-cn0-table1-version1 from allocator or local cache, then
   sends a lock request to cn0.
5. cn0 receive the lock request, but the lock-table bind is t1-cn0-table1-version2, and cn0 cn0 will consider
   this lock-table bind to be a remote lock table, because the serviceID(t1-cn0) != serviceID(t2-cn0). This
   will make rpc handle blocked.