Log Backup: Advancer cannot resolve lock when memory lock is not nil.

[3pointer]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

To reproduce this issue:

1. In a log backup cluster, ensure there is an existing lock and trigger the advancer to resolve the lock on specific regions.
2. Ensure the region contains a new lock ts in memory(from another new transaction).

2. What did you expect to see? (Required)

Resolve Lock success

3. What did you see instead (Required)

Resolve Lock Failed, wait for next round.

resolve locks failed, wait for next tick"",""category"":""advancer"",""uuid"":""log backup advancer"",""error"":""unexpected scanlock error: error:<locked:<primary_lock:xxx

4. What is your TiDB version? (Required)

master

[3pointer]

TiKV has an assumption: when any lock exists in memory less than max_ts, report Key is locked.

https://github.com/tikv/tikv/blob/37dfba1b046ebf308db972af0a3a788b7132244c/src/storage/mod.rs#L1634-L1641

We may need to adjust max_ts in the client from MaxUint64 to an appropriate range—greater than the lock ts but as small as feasible.

https://github.com/pingcap/tidb/blob/8a62d5a7bce32e56e22a8d17e85a086a2ceb1fd2/br/pkg/streamhelper/advancer.go#L691C17-L691C31