Provide color map for mask set with flag_values

[forman]

Is your feature request related to a problem? Please describe.

For dataset variables that provide a flag_values attribute, it would be convenient to get a matching matplotlib color mapping. For some variables we also provide flag_colors that should be used if so.

Describe the solution you'd like

Add method get_cmap() to xcube.core.maskset.MaskSet.

Additional context

CCI Land Cover datasets, lccs_class variable.

[konstntokas]

Describe the bug
MaskSet.get_cmap() method does not work, which has been added to xcube in #1012.

To Reproduce
In DeepESDL, use the CCI C3S land cover classification, create a mask and try to create a cmap. It creates a matplotlib.colors.ListedColormapinstance, and therefore the unit tests are working fine. If one wants to use the cmap, the following error is returned.
ValueError: data mapping points must start with x=0 and end with x=1

Additional context
The values of the colormap needs to be normalized. Furthermore, no color flag is given for 'no_data'. Here is a workable example, where we address the two issues.

def get_cmap(maskset):
    vals = maskset._flag_values
    extent = vals.max() - vals.min()
    vals_norm = (vals - vals.min()) / extent
    colors = ["#000000"] + maskset._flag_colors
    color_mapping = [(v, c) for v, c in zip(vals_norm, colors)]
    return matplotlib.colors.LinearSegmentedColormap.from_list(
        str(maskset._flag_var.name), color_mapping, N=len(vals)
    )

[forman]

PR underway...