Allow continuation

[jonthegeek]

Set up an option in {tb} to pass in a trained {tb} model as a starting point. {tabnet} has something like that. That way, if you’ve trained for 5 epochs and then decide you want to keep going, you could do so. We just have to check to make sure that the model has all the proper {tb} structure.

And, of course, that would let us test some things hackily, since we could create things that LOOK like they were trained via {tb}.