feat: update mlflow-related metadata models

[yoonhyejin]

Checklist

• The PR conforms to DataHub's Contributing Guideline (particularly Commit Message Format)
• Links to related issues (if applicable)
• Tests for the changes have been added/updated (if applicable)
• Docs related to the changes have been added/updated (if applicable). If a new feature has been added a Usage Guide has been added for the same.
• For any breaking change/potential downtime/deprecation/big changes an entry has been made in Updating DataHub

[jjoyce0510]

We should add a name field to mlModelGroup

[jjoyce0510]

ML Model Mapper + ML Model Group Mapper need to be updated to map the field "name"

[jjoyce0510]

Biggest modeling thing catching me up:

Data Process Instance + Data Job Entities already have an "outputs" lineage model. Simply extending them to include ML Model + ML Model Group is sufficient to render the lineage downstream (for DataJobs) and to show outputs of a Data Process Instance.

By starting to use the new training jobs field more extensively, we are introducing a large inconsistency in the model, a diverge between how lineage is specified for Datasets and AI Models that is sure to confuse our end users going forward.

I'd personally prefer to consider deprecating the trainingJobs field altogether and moving towards a model where the DataJob specifies it's outputs (as it already does with Datasets).

In the interim, we can support an either / OR type of situation at least between Data Job + ML Model / ML Model Group because it has traditionally already existed. But for Data Process Instance, I'd like to avoid introducing a new convention. Otherwise, we'll be forced to update the "outputs" of the Data Process rendering to fetch multiple different relationship types