You included data lineage tracking in the first part which probably needs to be piped from an orchestrator.

At this point it’s a build or buy type deal for models using a model registry service.

Data versioning still feels unsolved to me.