Adds data_source argument to covering grid

[chrishavlin]

This adds a data_source keyword (only) argument to the YTCoveringGrid and then intersects that data_source with the region used for the covering grid in order to create a regularized sampling of, e.g., a sphere.

Here's an example:

import yt
import numpy as np
import matplotlib.pyplot as plt 

ds = yt.load_sample("IsolatedGalaxy")

cg = ds.covering_grid(2, left_edge=[0.0, 0.0, 0.0], dims=[128, 128, 128])
gridded_dens = cg[("gas", "density")]

sp = ds.sphere(ds.domain_center, 0.2)
cg_sp_intx = ds.covering_grid(2, left_edge=[0.0, 0.0, 0.0], dims=[128, 128, 128], data_source=sp)
gridded_dens_in_sphere = cg_sp_intx[("gas", "density")]

f, axs = plt.subplots(nrows=1, ncols=2)
axs[0].imshow(np.log10(gridded_dens[:,:,64]))
axs[1].imshow(np.log10(gridded_dens_in_sphere[:,:,64]))

Pushing a nominally working draft now... but when ready this would close #4052

[chrishavlin]

note on current state: no new tests, a new example in the documentation would be good.

[chrishavlin]

edit: added tests and updated docs. Going to switch to ready-to-review. I won't be around for a little while but feel free to review!

[neutrinoceros]

refreshing CI. I will review this over the weekend

[neutrinoceros]

@yt-fido test this please

[neutrinoceros]

@yt-fido test this please

[neutrinoceros]

Nothing major, but I do have a couple requests. Overall, I really like this.

[neutrinoceros]

I see @matthewturk previously approved this but I'll hold off merging until he signs off again.