Support for matplotlib mosaic using variable names

Is your feature request related to a problem?

This is not related to any problem, but I think it would be nice to have a support for giving a matplotlib mosaic with the keys for the variables you want to plot for different panels and xarray parse that into the figure.

Describe the solution you'd like

Something like

import matplotlib.pyplot as plt
import xarray as xr
import numpy as np

n = 200
t = np.linspace(0,3*2*np.pi,n)
ds = xr.Dataset({letter:(("s","t"),np.sin(t)+0.5*np.random.randn(3,n)) for letter in "A B C D E".split()})
ds = ds.assign_coords(t=t,s=range(3))

mosaic = [
    ["A","A","B","B","C","C"],
    ["X","D","D","E","E","X"],
]

kw = dict(x="t",hue="s",add_legend=False)
ds.plot.line(mosaic=mosaic,empty_sentinel="X",**kw)

Describe alternatives you've considered

I have a code snippet that generate similar results but with more code.

import matplotlib.pyplot as plt
import xarray as xr
import numpy as np

n = 200
t = np.linspace(0,3*2*np.pi,n)
ds = xr.Dataset({letter:(("s","t"),np.sin(t)+0.5*np.random.randn(3,n)) for letter in "A B C D E".split()})
ds = ds.assign_coords(t=t,s=range(3))


mosaic = [
    ["A","A","B","B","C","C"],
    ["X","D","D","E","E","X"],
]

kw = dict(x="t",hue="s",add_legend=False)
fig = plt.figure(constrained_layout=True,figsize=(8,4))
ax = fig.subplot_mosaic(mosaic,empty_sentinel="X")
for key in ds:
    ds[key].plot.line(ax=ax[key],**kw)

Additional context

No response

Start with the ds.plot.line entry point and compare the requested API with matplotlib's Figure.subplot_mosaic behavior shown in the examples. Check how dataset variables are mapped to axes and how empty_sentinel is handled. Done means a mosaic can route each named variable to its panel while preserving the existing line-plot options and behavior.