Unstable test ReplicaFetcherThreadTest#testFollowerHighWatermarkHigherThanOrEqualToLeader

### Search before asking

- [x] I searched in the [issues](https://github.com/alibaba/fluss/issues) and found nothing similar.

### Fluss version

main (development)

### Please describe the bug 🐞

https://github.com/alibaba/fluss/actions/runs/14530474925/job/40769262274

```
[INFO]
Error: Failures:
Error: ReplicaFetcherThreadTest.testFollowerHighWatermarkHigherThanOrEqualToLeader:201
Expecting actual:
7430L
to be greater than or equal to:
7440L

[INFO]
Error: Tests run: 402, Failures: 1, Errors: 0, Skipped: 0
```

The test was introduced in https://github.com/alibaba/fluss/pull/676

### Solution

_No response_

### Are you willing to submit a PR?

- [ ] I'm willing to submit a PR!

Start with ReplicaFetcherThreadTest#testFollowerHighWatermarkHigherThanOrEqualToLeader at line 201 and review the test's introduction in pull request #676. Reproduce the failure from the linked CI run, then trace why the follower high watermark can be below the leader's; done means the test passes reliably without masking the replication condition.

Written by the indexing model from the issue text.