matplotlib compatibility issues

[kecnry]

• matplotlib 3.5.2 breaks colorbars/sidebars with the following traceback (see draw_sidebars = True returns error when plotting mesh #990):

  1446 fig = parents[0].get_figure()
   1448 pad0 = 0.05 if fig.get_constrained_layout() else loc_settings['pad']
   1449 pad = kw.pop('pad', pad0)

AttributeError: 'tuple' object has no attribute 'get_figure'

• matplotlib 3.6 with a GTK backend breaks default plotting cases (see Line profile bugfix (2.4.7 release) #637)

• matplotlib 3.7.2 (or sooner) breaks when attempting to draw axvlines (see solver tutorial #881).

[MichaelAbdul-Masih]

When trying to run

b.plot(solution='lcgeom_solution', show=True)

after running the LC geometry estimator (from the 2023 workshop tutorial(12), an error is thrown and the eclipse edges are not plotted.

numpy.__version__ = 1.24.3
matplotlib.__version__ = 3.7.1

EDIT: it appears that downgrading to matplotlib 3.5.1 does not work, but downgrading numpy to 1.23.1 does solve the issue so perhaps this is a numpy issue.