Added functionality to model.py and opnmf.py to run out-of-sample OPNMF.

[Alfiew]

Created a 'transform' function, to enable running out-of-sample OPNMF.

estimator = opnmf.model.OPNMF(n_components=n)

As before, to fit and transform a given dataset

W = estimator.fit_transform(data)

New addition, to then transform a different dataset (same structure)

H, mse = estimator.transform(new_data)

[fraimondo]

Hi @Alfiew !

Thanks for this contribution. I was actually working on this myself last week and did not manage to finish it.

Can you add a very simple test?

The idea is that a fit_transform(X) call should output the same as .fit(X).transform(X) call.