Add the ability to use a FITS skymap as input to pycbc_make_sky_grid

[titodalcanton]

After #4267 and #4984, the next and last step for using HEALPix skymaps as input to PyGRB is to generate a sky grid that is constrained by a HEALPix skymap.

Some considerations to drive the design of the CLI interface:

• Well-formed skymaps should contain the trigger time in their DATEOBS header field, so in principle the FITS file is all that should be needed to specify both the time and sky location.
• It would be good to keep the ability to use the existing CLI arguments (central sky location + radius) as well, in case we want to follow up external triggers that do not provide HEALPix skymaps.
• It would be good to have the ability to construct an all-sky grid, for experimenting with all-sky coherent searches.

In terms of the overall code (re)design, it would be good (if possible) to maintain a certain separation (i.e. modularity) between the various possible constraints on the sky location (no constraint = all-sky grid, constraint based on center and radius, constraint based on a HEALPix skymap) and the various possible ways to define the density of grid points (constant density, density defined by timing considerations only, density defined by fractional loss of incoherent SNR, density defined by fractional loss of coherent SNR, exact SNR loss vs metric approximation, etc).