Applying Model instances to db

There are 4 key AI Model sub classes, see here for detailed usage:

Path	Description
superduper.components.model.ObjectModel	Wraps a Python object to compute outputs
superduper.components.model.APIModel	Wraps a model hosted behind an API to compute outputs
superduper.components.model.QueryModel	Maps a Database select query with a free variable over inputs
superduper.components.model.SequentialModel	Computes outputs sequentially for a sequence of Model instances

As well as these key sub-classes, we have classes in the superduper.ext.* subpackages: See here for more information.

Whenever one of these Model descendants is instantiated, and db.apply(model) is called, several things can (do) happen:

1. The Model's metadata is saved in the db.metadata_store.
2. It's associated data (e.g.) model is saved in the db.artifact_store.
3. (Optional) if the Model has a Trainer attached, then the Model is trained/ fit on the specified data.
4. (Optional) if the Model has an Evaluation method attached, then the Model is evaluated on the specified data.