Accept numpy arrays (not lists) to ensemble.train

[anthonymorast]

If all of the ensemble members should be trained on the same training data the training method should be able to take only the training x and y data rather than lists of the same training and test data. This would reduce confusion and memory footprint.

That is,

ensemble.train([trainx for _ in range(nummodels)], [trainy for in range(num_models)])

would become

ensemble.train(trainx, trainy)

[anthonymorast]

This would also prevent having to update the num_models variable in the train call when adding new models to the ensemble as all members would just use the same passed data.