pipeline component not updating or creating a new version in the registry.

### Operating System

Linux

### Version Information

latest azure python sdk for python 3.7

### Steps to reproduce

If I create a pipeline with the decorator and deploy it to a batch endpoint, Azure ML will automatically create a pipeline component in the registry. Sounds great right? Not so much if you want to CHANGE such a pipeline. Suppose I change my pipeline but keep the same name. The new pipeline will test ok, but when I attempt to update the batch endpoint the modifications are not being put in place!

Workaround: Create everything new with new names (pipeline, endpoint, and deployment).

Also, there is no way to delete the pipeline component from the ML studio. That feature should exist since we can delete the endpoint and deployments.

Thanks

Arthur

### Expected behavior

It is expected that a new version of the pipeline component will be created in the registry.

### Actual behavior

The pipeline component is not updated.

### Addition information

_No response_

No repository file, test, or entry point is identified in the report. Start by reproducing the decorator-based pipeline deployment to a batch endpoint with the Azure Python SDK, then compare registry behavior after changing the pipeline while keeping its name; done means the changed pipeline is deployed as a new registry version and the deletion behavior is clarified.

Written by the indexing model from the issue text.