Attach custom beam

[e-koch]

From #432. A beam object can now be attached with:

newcube = cube.with_beam(beam)
newproj = projection.with_beam(beam)

Since radio-beam is now a dependency, I've removed the read_beam option from SpectralCube. We need to keep it for Projection though since there remains the issue with non-matching celestial axes for some slices.

[coveralls]

Coverage increased (+0.1%) to 74.941% when pulling e5ff430 on e-koch:attach_custom_beam into 668e322 on radio-astro-tools:master.

[keflavich]

I would prefer this be a with_beam method; nearly everything else we do related to modifying cubes forces the creation of a new cube object, and it feels weird having a one-off attach method. If a beam is really needed as part of the original cube, it should be established during instantiation.

[keflavich]

we're already in the else of if beam is None, so this if could just be deleted with no change of indentation, I think

[e-koch]

no, I'll have to duplicate the three lines below that if without it for when the beam is loaded from the header. The other option would be to move loading a beam from the header somewhere into io.fits

[keflavich]

right, good point, OK.

[e-koch]

Agreed. I changed both to be with_beam and return a new object.

It highlighted one issue in 2D objects when the slice is taken along a position-velocity slice:

___________________________ TestSpectralCube.test_getitem[vda_beams-trans8] ____________________________

self = <spectral_cube.tests.test_spectral_cube.TestSpectralCube object at 0x7f33e9101190>
name = 'vda_beams', trans = [0, 2, 1]

    @pytest.mark.parametrize(('name', 'trans'), translist)
    def test_getitem(self, name, trans):
        c, d = cube_and_raw(name + '.fits')
    
        expected = np.squeeze(d.transpose(trans))
    
        assert_allclose(c[0,:,:].value, expected[0,:,:])
>       assert_allclose(c[:,:,0].value, expected[:,:,0])

spectral_cube/tests/test_spectral_cube.py:257: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
spectral_cube/spectral_cube.py:2862: in __getitem__
    meta=meta)
spectral_cube/lower_dimensional_structures.py:236: in __new__
    self.beam = beam
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = <Slice [[ 0.54671028, 0.96958463, 0.93949894],
        [ 0.59789998, 0.0884925...2873751, 0.28093451],
        [ 0.14092422, 0.07455064, 0.77224477]] Jy / beam>
obj = [Beam: BMAJ=0.10000000149 arcsec BMIN=0.40000000596 arcsec BPA=0.0 deg, Beam: BMAJ=0.20000000298 arcsec BMIN=0.3000000... arcsec BMIN=0.20000000298 arcsec BPA=60.0 deg, Beam: BMAJ=0.40000000596 arcsec BMIN=0.10000000149 arcsec BPA=30.0 deg]

    @beam.setter
    def beam(self, obj):
    
        if not isinstance(obj, Beam):
>           raise TypeError("beam must be a radio_beam.Beam object.")
E           TypeError: beam must be a radio_beam.Beam object.

We should probably just not allow this type of slicing with a VRSC.

[e-koch]

@keflavich - separate PR for the above error? One test should fail on travis.

[keflavich]

It's OK to fix it here if you want

[coveralls]

Coverage increased (+0.2%) to 74.985% when pulling a697f51 on e-koch:attach_custom_beam into 668e322 on radio-astro-tools:master.

[coveralls]

Coverage increased (+0.1%) to 74.964% when pulling ee56618 on e-koch:attach_custom_beam into 668e322 on radio-astro-tools:master.

[coveralls]

Coverage increased (+0.1%) to 74.964% when pulling cca53b8 on e-koch:attach_custom_beam into 668e322 on radio-astro-tools:master.