DOC: Update some plotting code to current Matplotlib idioms

[timhoffm]

• 3D Axes are created via add_subplot(projection='3d')
• There is now a stairs() function that's specifically designed for
  showing histogram curves
• Labels should be passed as keyword arguments to the plot functions
  instead of to legend(), which reduces the risk of mixing them up.

[rgommers]

This looks great, thanks @timhoffm. One issue, the build is failing with:

AttributeError: module 'matplotlib.pyplot' has no attribute 'stairs'

How recently was stairs introduced? We can consider upgrading the minimum matplotlib version needed to build the docs. For that, stairs has to be in a release that's >=1 year old I'd say. Otherwise it's better to revert that change.

[charris]

stairs was new in v3.4.0, might could be done with steps if v3.4.0 is too new.

[timhoffm]

I've reverted the use of stairs. Other ways to plot histograms from binned data are cumbersome/unituitive. That would be either using bars

>>> centers = (bins[1:] + bins[:-1]) / 2
>>> widths bins[1:] - bins[:-1]
>>> plt.bar(centers, n, width=widths)

or step (with the problem that steps expects equal-length inputs):

>>> plt.step(bins, np.insert(n, 0, [0]))

That's exactly why we've introduced stairs. I don't want to advertise any of the workarounds. So, I'm going back to the original plot. I've added a note that plt.stairs can be used for Matplotlib 3.4+. This way people can still learn the new function and it's more likely that somebody will remove plot from the docs some time in the future.

[rgommers]

LGTM now, thanks @timhoffm