Add a check in cdf_to_xarray to ensure DEPEND_X types are string

[bryan-harter]

In:

def _discover_depend_variables(vardata: Dict[str, npt.NDArray], varatts: Dict[str, Any], varprops: Dict[str, VDRInfo]) -> List[str]:
    # This loops through the variable attributes to discover which variables are the coordinates of other variables,
    # Unfortunately, there is no easy way to tell this by looking at the variable ITSELF,
    # you need to look at all variables and see if one points to it.

    depend_regex = re.compile("depend_[0-9]+$")

    list_of_depend_vars = []

    for v in varatts:
        depend_keys = [x for x in list(varatts[v].keys()) if depend_regex.match(x.lower())]
        for d in depend_keys:
            if varatts[v][d] in vardata:

It'll error out if the Depend_X attribute is anything other than a string. So first we should add a check that "d" is a string, and give an error if not.

Then we should write a unit test, and ensure it fails