An adapter to run a single model

[ardunn]

Useful for retaining the other parts of the ml pipeline without running tpot.

This adapter's init should take in a model object implementing fit and predict methods (or if you can figure out a way to do it with models not having these methods, that's fine too) and running it should fit/predict that single model (e.g., RandomForestRegressor) in a way analogous to TPOTAdaptor.

[ardunn]

done via 8cb2b27 + commits soon after

@utf might be interested in this