Helper function for accessing the `xarray.DataArray` of a `MultiscaleSpatialImage`

FYI @giovp

As explained here https://github.com/scverse/spatialdata/issues/115, we assume that for each level of a MultiscaleSpatialImage we have only one variable. So, it would be handy to have a convenience function to get directly the xarray.DataArray for a particular multiscale level.

This could be a method of MultiscaleSpatialImage that we could propose to add upstream in the multiscale-spatial-image library, or a convenience function like the following, to define in our repo.

def get_scale(multiscale_image: MultiscaleSpatialImage, level: int) -> xarray.DataArray:
    # code is explained in the examples below
    d = dict(multiscale_image[f'scale{level}'])
    assert len(d) == 1
    return d.values().__iter__().__next__()

At the moment I am using repeatedly the following code (for instance to iterate over the various levels of the pyramid):

multiscale_image # a MultiscaleSpatialImage object
for level in multiscale_image.keys():
    node = multiscale_image[level]
    d = dict(node)
    assert len(d) == 1
    xarray_data = d.values().__iter__().__next__()

an alternative approach is the following (but it is also verbose):

multiscale_image # a MultiscaleSpatialImage object
for level in multiscale_image.keys():
    node = multiscale_image[level]
    variables = node.variables
    assert len(variables) == 1
    xarray_data = d[variables[0]]

Start by reviewing the MultiscaleSpatialImage usage and the helper examples in this issue, then inspect the repository for existing multiscale image utilities. Decide whether the convenience function belongs in this repository or upstream. Done means a callable helper or method returns the single xarray.DataArray for a requested multiscale level, with coverage for the expected one-variable assumption.