Enable integration with the mlflow registry

[Galileo-Galilei]

Description

This is a catch-all ticket for model registry integration. Some ideas:

• Add a key in mlflow.yml to set mlflow_registry_uri
• Create a dataset which is able to save/load a specific version of a model in the registry uri
• Add the possibility to create a new version of the same model instead of a new run when running e PipelineML with pipeline_ml_factory

Context

The model registry is becoming more and more prevalent while deploying in production and there is currently o support for this mlflow feature.

Possible Implementation

TBD