use matplotlib's categorical axis features

[dcherian]

Is your feature request related to a problem? Please describe.
xarray currently doesn't allow plotting against coordinates with string labels for example.

Describe the solution you'd like
Use matplotlib's categorical axis support. Example: https://matplotlib.org/gallery/lines_bars_and_markers/categorical_variables.html

This may be the only place a change is required:

xarray/xarray/plot/utils.py

Lines 572 to 608 in 4e89331

	def _ensure_plottable(*args):
	"""
	Raise exception if there is anything in args that can't be plotted on an
	axis by matplotlib.
	"""
	numpy_types = [np.floating, np.integer, np.timedelta64, np.datetime64, np.bool_]
	other_types = [datetime]
	try:
	import cftime
	cftime_datetime = [cftime.datetime]
	except ImportError:
	cftime_datetime = []
	other_types = other_types + cftime_datetime
	for x in args:
	if not (
	_valid_numpy_subdtype(np.array(x), numpy_types)
	or _valid_other_type(np.array(x), other_types)
	):
	raise TypeError(
	"Plotting requires coordinates to be numeric, boolean, "
	"or dates of type numpy.datetime64, "
	"datetime.datetime, cftime.datetime or "
	f"pandas.Interval. Received data of type {np.array(x).dtype} instead."
	)
	if (
	_valid_other_type(np.array(x), cftime_datetime)
	and not nc_time_axis_available
	):
	raise ImportError(
	"Plotting of arrays of cftime.datetime "
	"objects or arrays indexed by "
	"cftime.datetime objects requires the "
	"optional `nc-time-axis` (v1.2.0 or later) "
	"package."
	)