Removing location kwarg from moon illumination/phase computations

[bmorris3]

This PR removes the location keyword argument from the moon_illumination, moon_phase_angle functions in the moon module.

The motivation here is that astropy's get_moon does not yet support computations for multiple times at a non-geocentric observer, so we were calculating each moon position with a slow for loop (astropy/astropy#5216). While the location of the observer on the surface of the Earth matters a lot if you want precise moon positions on the celestial sphere, it doesn't affect the result much for computing moon phase/illumination compared to a geocentric observer. So we can get vast speed-ups if we just drop the location specification here, where it doesn't matter much.

I've simply removed the location keyword argument, which now computes all phases/illuminations from GCRS=(0, 0, 0).

cc @StuartLittlefair @eteq @kvyh

[StuartLittlefair]

HI @bmorris3

I've left a comment on astropy/astropy#5216, but there's a PR for astropy which will fix that issue. That PR has a milestone of v1.2.2. I guess you might want to merge this anyway if you want the next astroplan release to support astropy 1.2 and 1.2.1...

[bmorris3]

I'm going to wait until at least tomorrow to merge. This should also help speed up your constraint computations when the moon illumination is evaluated.

[jcurbo]

Hi, it looks like when you removed the location requirement you didn't update the docstrings, so the generated docs (see here) still say you need to include a location. Looks like a straightforward change (remove two lines).

Also the same for moon_phase_angle.

[bsipocz]

@jcurbo - Thanks for reporting this. #284 addresses the issue.