Update API documentation and switch to @properties for some Descriptor attributes.

[andrewdnolan]

This PR:

1. Improves the documentation, switch to numpy style docstrings. The primary improvements are made for the Descriptor class, which has improved formatting and more detailed information (for both the class itself and it's attributes).

2. I've switched the projection and latlon attributes to be properties. I did this in order to move some logic outside of the method construct to abstract things a little. Most importantly the coordinate arrays (from the minimal dataset) transformation has been moved to the Descriptor.projection setter method so that reprojection to a new map projection after Descriptor initialization is now possible, although not really advised. These abstraction should also help set the groundwork for planar-periodic mesh support.

[xylar]

@andrewdnolan, these changes look great to me! I have some picky suggestions about single-line docstrings below.

I built the docs locally and looked through them. Everything looked as expected -- very clean!

[andrewdnolan]

Testing

Here's a demonstration of the reprojection capabilities enable by the Descriptor.projection setter method:

import cartopy.crs as ccrs
import mosaic
import matplotlib.pyplot as plt
import xarray as xr

def make_figure(ds, descriptor):

    fig, ax = plt.subplots(
        figsize=(9,7), constrained_layout=True, subplot_kw=dict(projection=descriptor.projection)
    )

    collection = mosaic.polypcolor(ax, descriptor, ds.indexToCellID, antialiaseds=True)

    ax.coastlines()
    fig.colorbar(collection, fraction=0.1, shrink=0.5, label="Cell Index");

    return fig, ax

ds = mosaic.datasets.open_dataset("QU.240km")

projection_one = ccrs.InterruptedGoodeHomolosine()
projection_two = ccrs.LambertConformal()
transform = ccrs.PlateCarree()

descriptor = mosaic.Descriptor(ds, projection_one, transform)

# first figure with original projection 
fig_one, ax_one = make_figure(ds, descriptor)

# switch the projection of the descriptor 
descriptor.projection = projection_two

# second figure with new projection
fig_two, ax_two = make_figure(ds, descriptor)

plt.show()

Figure One:

Figure Two: