Adding a colorbar with matplotlib backend moves the x/y axes into the center of the plot

[stefmolin]

ALL software version info

(this library, plus any other relevant software, e.g. bokeh, python, notebook, OS, browser, etc)

• Python 3.8
• jupyterlab 3.1.1
• geoviews 1.9.1
• holoviews 1.14.5
• matplotlib 3.4.3

Description of expected behavior and the observed behavior

Adding the colorbar should not affect the x/y axes of the plot (world map), but it is moving both to the center for some reason.

Complete, minimal, self-contained example code that reproduces the issue

import geoviews as gv
import geoviews.feature as gf
gv.extension('matplotlib')

(gf.land * gf.coastline * gf.borders * gv.Points(
    [{'latitude': 0, 'longitude': 0, 'mag': 7},
     {'latitude': 10, 'longitude': 10, 'mag': 2},
     {'latitude': -10, 'longitude': -10, 'mag': 0}], kdims=['longitude', 'latitude'], vdims='mag'
).opts(color='mag', cmap='fire', colorbar=True)).opts(global_extent=True)

Screenshots or screencasts of the bug in action

[jbednar]

Yikes! I can reproduce that with geoviews=1.9.2. holoviews=1.14.6, and matplotlib=3.4.3. Hard to see how that could be HoloViews rather than Matplotlib itself, but I'm not sure how to reproduce that with pure Matplotlib.