Making a pb mask requires hard-coding units that are dimensionless

[abulatek]

Ran into an issue when trying to work with a primary beam file in spectral-cube. File was imported as follows:

import glob
from spectral_cube import SpectralCube
pb_fn = glob.glob(f"primary_beam.pb") # any .pb file
pb = SpectralCube.read(pb_fn[0], format = 'casa_image')

Running pb.unit gives a blank cell, confirming that that cube is unitless. However, trying to mask above a certain primary beam level yields an error.

mask = pb > 0.8

ValueError                                Traceback (most recent call last)
<ipython-input-26-6ec680c9e817> in <module>()
----> 1 mask = pb > 0.8

/blue/adamginsburg/abulatek/spectral-cube/spectral_cube/spectral_cube.py in __gt__(self, value)
   2235             The threshold
   2236         """
-> 2237         value = self._val_to_own_unit(value)
   2238         return LazyComparisonMask(operator.gt, value, data=self._data, wcs=self._wcs)
   2239 

/blue/adamginsburg/abulatek/spectral-cube/spectral_cube/spectral_cube.py in _val_to_own_unit(self, value, operation, tofrom, keepunit)
   2224                              " SpectralCubes or Quantities with "
   2225                              "a unit attribute."
-> 2226                              .format(operation=operation, tofrom=tofrom))
   2227 
   2228     def __gt__(self, value):

ValueError: Can only compare cube objects to SpectralCubes or Quantities with a unit attribute.

Since pb.unit is dimensionless, this should work. What we have to do instead is:

mask = pb > 0.8 * pb.unit

[keflavich]

The fundamental problem here is that we should let u.dimensionless_unscaled go through w/o attempting to convert.

[e-koch]

Fixed in #820. Thanks @abulatek!