BUG: mask_image should apply the mask to the input

[cookpa]

The docstring said:

level : scalar or tuple of scalars
Level(s) at which to mask image. If vector or list of values, output image is non-zero at all locations where label image matches any of the levels specified.

But this is only true if binarize is True.

The binarize option

binarize : boolean
whether binarize the output image

didn't actually binarize the output, it binarized the mask without reference to the scalar image. So

>>> masked = ants.mask_image(img, mask) # NOT non-zero where mask matches level
>>> masked_binary = ants.mask_image(img, mask, binarize=True) # NOT a binarized version of masked

Now fixed to always mask the scalar image, and binarize afterwards if requested. This can be used to subset labels with

ants.mask_image(labels,labels, level=(1,2,3,4))

or create a binary mask

ants.mask_image(labels,labels, level=(1,2,3,4), binarize=True)

[coveralls]

coverage: 76.915% (+0.01%) from 76.901%
when pulling de7b650 on mask_binarize
into c8b0add on master.

[cookpa]

@stnava @ntustison any objections to this fix? I looked at antspynet and antsMM, and I don't see any uses that would be affected.

[ntustison]

@cookpa Yeah, fine with me. I didn't write this and can't remember anytime I've ever used it.