Update Facetgrid api to match seaborn

Something I assumed would work:

g = ds.plot(hue="level", col="time")
# Second plot uses the existing FacetGrid's axes
ds2.plot(ax=g.axes.flat, hue="level", col="time")

Gives ValueError: Can't use axes when making faceted plots.

How I currectly do it:

Xarray's plotting with facets doesn't work the same way as seaborn's FacetGrid. You currently need to use a manual approach, such as :

g = ds.plot(hue="level", col="time")
for i in range(ds2.time.size):
    ax = g.axs.flat[i]
    ax.set_prop_cycle(None)
    ds2.isel(time=i).plot(ax=ax, hue="level")

More generally:

I think it would be nice to have the functionalities seaborn's facetgrid has, notably passing a facetgrid to the plotting function directly but also including the utility functions that handle the legend or labels to be able to have nice better plots without changing the datasets themselves.

Start by reviewing xarray's existing faceted plotting API and seaborn's FacetGrid behavior, especially how axes, legends, and labels are handled. Done would include support for passing an existing FacetGrid and the requested plotting utilities, but the issue does not identify files or tests.