bookkeeper storage ledgers did not auto delete [Bug]

### Search before asking

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

### Version

System: Centos7.6
Pulsar: 2.8.2

### Minimal reproduce step

All the namespaces/topics we set retention policy and ttl.
During pulsar manager we can see all namespace take about 25G storage size：

But in system it used nearly 100G

I don't know why, i already use "curl -XPUT 127.0.0.1:8888/api/v1/bookie/gc" try release disk space and it not work.

### What did you expect to see?

Disk usage similar with what i see in pulsar manage.

### What did you see instead?

about 90G instead of 25G

### Anything else?

_No response_

### Are you willing to submit a PR?

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

Start by reproducing the storage discrepancy with namespace retention and TTL enabled, then investigate the bookie GC entry point at /api/v1/bookie/gc and compare its result with the Pulsar Manager usage. Done means expired ledger data is removed and system disk usage is consistent with the reported namespace storage.

Written by the indexing model from the issue text.