[REVIEW] Added ability to call cuML with numpy arrays and general test cleaning

[dantegd]

PR adds the ability to call cuML methods using numpy arrays (except for knn for now):

from cuml import TruncatedSVD as cuTSVD
cutsvd = cuTSVD(n_components=1)
X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [2, 1], [3, 2]],
                     dtype=np.float32)
X_t = cutsvd.fit_transform(X)

Also major cleaning of the python tests and expanded coverage. Upcoming PR will clean and improve kmeans, and overall improve coverage.

Also solves bugs dbscan and other methods.

[Iroy30]

Would be nice to return self in fit to comply with sklearn like api

[oyilmaz-nvidia]

@dantegd Just tested and it worked for me. Great work! Quick question. Can we add these numpy examples in the jupyter notebooks? I think people will like to see these code examples. Maybe Jiwei can help on this. Any thoughts?