[Bug] Synchronization using geo's namespace-level replicas Message lost

### Search before reporting

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

### Read release policy

- [x] I understand that [unsupported versions](https://pulsar.apache.org/contribute/release-policy/#supported-versions) don't get bug fixes. I will attempt to reproduce the issue on a supported version of Pulsar client and Pulsar broker.

### User environment

pulsar version 3.0.11

### Issue Description

Currently, there are two clusters that have set up geo namespace replica synchronization, but in one of the clusters, it is found that the messages synchronized from the cluster are far different from the main cluster traffic and storage.

### Error messages

```text
No error message was seen
```

### Reproducing the issue





### Additional information

null

### Are you willing to submit a PR?

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

Start by reproducing the geo namespace replica synchronization issue with Pulsar 3.0.11, or a supported client and broker version, using the two-cluster setup described. Compare synchronized message traffic and storage between clusters; done means messages are no longer lost and replica traffic matches the main cluster.

Written by the indexing model from the issue text.