How can the trained model be saved for future scoring?

You can use the 'store' parameter to save the model in a binary table object. This object can then be used for scoring new data with the bartScore action.