[FEA]: Replace the tensor copy in WriteToVectorDBStage with inserting AddScoresStage to the pipeline

[dagardner-nv]

Is this a new feature, an improvement, or a change to existing functionality?

Change

How would you describe the priority of this feature request

Low (would be nice)

Please provide a clear description of problem this feature solves

WriteToVectorDBStage has some logic in it, that if the incoming message is of type MultiResponseMessage it will copy the probs tensor into the dataframe in the embedding_column_name parameter.
https://github.com/nv-morpheus/Morpheus/blob/fea-sherlock/morpheus/stages/output/write_to_vector_db_stage.py#L135

This is also a bit tricky as the embedding_column_name parameter is only ever used when the incoming message type is a MultiResponseMessage instance and ignored otherwise, however this isn't reflected in the docstrigns.

Describe your ideal solution

The AddScoresStage should be used for this purpose instead as it contains a C++ impl, and ensures that the destination column will be pre-allocated.

Additional context

No response

Code of Conduct

• I agree to follow this project's Code of Conduct
• I have searched the open feature requests and have found no duplicates for this feature request

[dagardner-nv]

This one is a bit trickier than at first glance, the AddScoresStage is tied to the config's classification labels, in addition to this, it copies each column in the probs tensor to a different column in the DF.

On the other hand the WriteToVectorDBStage is expecting the probs tensor to be copied into a single array column.