Possible update in NumPy triggering warning for NaN addition

[rhugonnet]

I've seen tests of volume.py fail several times in the past days due to a RuntimeWarning triggered during NaN addition that we use to initiate our empty arrays (in #227).
For example in line 552 of volume.py:

    values = np.empty((n_bins, unique_indices.shape[0]), dtype=float) + np.nan

The failure was a bit random, so I guess it depends on the version of numpy set up in the environment, and this warning might be a recent addition. If this is the case (I couldn't find confirmation with a quick look), we might have to update a lot of initializations!