Projection mismatch after library upgrade

[valenal]

when upgrading matplotlib-base from 3.2.1 --> 3.4.1
and using ds.monet.quick_map(robust=True)

This is what my code looks like (the only difference between the attached figures is the library)

import monet as m
import monetio as mio
ds = mio.cmaq.open_dataset('CCTM_ASENS_v52_DDM_gcc_DDM_P2_test_TRECH2_20160624.nc')
print(repr(ds[poll][15,0,160:163,178:181]))
ds[poll][15,0,160:163,178:181].monet.quick_map(robust=True)

<xarray.DataArray 'O3_PNX' (y: 3, x: 3)>
array([[-1.413654 , -1.2353712 , -0.93085957],
[-1.1120044 , -1.2926427 , -1.1387304 ],
[-0.70462906, -0.980141 , -1.0245911 ]], dtype=float32)
Coordinates:
time datetime64[ns] 2016-06-24T15:00:00
longitude (y, x) float64 -74.17 -74.03 -73.89 ... -74.09 -73.96 -73.82
latitude (y, x) float64 40.66 40.63 40.6 40.76 ... 40.71 40.87 40.84 40.81
Dimensions without coordinates: y, x
Attributes:
long_name: O3_PNX
units: ppbV
var_desc: Sens of O3 to PNX
proj4_srs: +proj=lcc +lat_1=33.0 +lat_2=45.0 +lat_0=40.0 +lon_0=-97.0 +x...

matplotlib-base 3.2.1

matplotlib-base 3.4.1

[zmoon]

In v3.3 Matplotlib changed the default for grid layout assumptions for pcolors: https://matplotlib.org/stable/users/prev_whats_new/whats_new_3.3.0.html#pcolor-and-pcolormesh-now-accept-shading-nearest-and-auto, that's what causes this change. But monet should probably address this explicitly (by setting Matplotlib version and/or providing shading when plotting).