[Enhancement] Not yet make full use of Polymorphism in Producer and AbstractProducer.

### Search before asking

- [X] I had searched in the [issues](https://github.com/apache/eventmesh/issues?q=is%3Aissue) and found no similar issues.

### Enhancement Request

* In storage modules of knative, rocketmq, pulsar, kafka, the model of `Producer`'s hierarchy is not unified and confused: 3 `AbstractProducer`s in module knative, rocketmq, pulsar have some similar contents, but some implement `Producer` and some not. Even `AbstractProducer` in kafka is not been designed into `Producer`'s hierarchy.
* `InterruptedException` should never be ignored. Or the import interrupted state of Thread may be ignored unexpected.
* Some logic has been provided by existing util, so no need to code again.
* Smaller `synchronized` scope, lower the thread competitiveness
* In AtomicXxx, `set()` is thread safe and better than CAS when no need to get a return value.

### Describe the solution you'd like

* Unify the model of `Producer`'s hierarchy: Rename 3 `AbstractProducer` in module knative, rocketmq to `KnativeAbstractProducer, RocketmqAbstractProducer`. Add a `AbstractProducer` in api module implements `Producer`. Make `AbstractProducer`s in module knative, kafka, rocketmq, pulsar all extends `AbstractProducer` in api module or its sub abstract class. Remove `AbstractProducer` in module pulsar.
* Catch `InterruptedException` and call `interrupt()` of current Thread.
* Use existing util.
* Minimize `synchronized` scope.
* Use `set` replace CAS where no need get a return value.

### Are you willing to submit PR?

- [X] Yes I am willing to submit a PR!

Start by comparing the Producer and AbstractProducer hierarchies in the api, knative, rocketmq, pulsar, and kafka storage modules. Trace the existing interruption handling, utility usage, synchronization, and AtomicXxx updates before deciding how the shared hierarchy should be structured. Done means the producer hierarchy is unified, interrupted state is preserved, and the listed concurrency cleanups are applied consistently.

Written by the indexing model from the issue text.