Netty globalEventExecutor leak classloader in flink catalog.

### Search before asking

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

### Fluss version

0.8.0 (latest release)

### Please describe the bug 🐞

Currently, flink gateway with fluss long time running, will thread leak. When dump the thread, there are lots of threads(named globalEventExecutor-2-2) left.
```shell
Thread/Slack: org.apache.fluss.shaded.netty4.io.netty.util.concurrent.FastThreadLocad @0001
name: globalEventExecutor-2-2
Thread/Slack: org.apache.fluss.shaded.netty4.io.netty.util.concurrent.FastThreadLocad @0002
name: globalEventExecutor-2-1
...... lot of ....
```

It seems the problem of https://github.com/netty/netty/issues/14615
> If Netty is on a main classpath in a system that creates ephemeral classloaders, e.g. for plug-ins, then we should avoid holding on to those ephemeral classloaders.

The fluss connector is in the main classpath, and gateway will create ephemeral classloaders.

### Solution

_No response_

### Are you willing to submit a PR?

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

Start by reproducing the long-running Flink gateway with Fluss 0.8.0 and inspect the leaked globalEventExecutor threads described in the report. Read the linked Netty issue and trace how the Fluss connector and gateway ephemeral classloaders interact; done means the gateway no longer accumulates these threads or retains their classloaders.

Written by the indexing model from the issue text.