Testing is far too fragile

The failures come from the image comparison in tests/integration/test_advanced_plot_examples.py and tests/integration/test_contour_plot_examples.py, which compare rendered plots against reference images in tests/reference-example-images. Run one failing case, such as test_example_1_basic_cylindrical, and read how the rms value is compared with the reference. The issue does not say whether the fix should be a tolerance change, regenerated references, or environment pinning, so a maintainer decision is needed first. Done means the suite passes across environments under the chosen approach.

Subject of the issue

I've just pulled the latest release, and 24 tests fail. They are all minor fails to do with image mismatch, but it's just too fragile.

================================================================= short test summary info ==================================================================
FAILED tests/integration/test_advanced_plot_examples.py::test_example_14_vector_with_colour_contour - Failed: Image mismatch for example 14: {'rms': np.float64(0.07727821087173495), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_advanced_plot_examples.py::test_example_15_polar_vector_plot - Failed: Image mismatch for example 15: {'rms': np.float64(1.2350842133667295), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference...
FAILED tests/integration/test_advanced_plot_examples.py::test_example_16b_basic_stream_plot - Failed: Image mismatch for example 16b: {'rms': np.float64(19.617677264399287), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_advanced_plot_examples.py::test_example_16c_enhanced_stream_plot - Failed: Image mismatch for example 16c: {'rms': np.float64(21.34025810212168), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference...
FAILED tests/integration/test_advanced_plot_examples.py::test_example_17_basic_stipple_plot - Failed: Image mismatch for example 17: {'rms': np.float64(0.21548615766477328), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_advanced_plot_examples.py::test_example_18_polar_stipple_plot - Failed: Image mismatch for example 18: {'rms': np.float64(0.14397116764560633), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_advanced_plot_examples.py::test_canari_1 - Failed: Image mismatch for example canari_1: {'rms': np.float64(0.02305860896416126), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/re...
FAILED tests/integration/test_advanced_plot_examples.py::test_example_40_tracks_polar_stereographic - Failed: Image mismatch for example 40: {'rms': np.float64(0.01662284425174423), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_contour_plot_examples.py::test_example_1_basic_cylindrical - Failed: Image mismatch for example 1: {'rms': np.float64(0.18352397386241887), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference...
FAILED tests/integration/test_contour_plot_examples.py::test_example_3_map_limits_and_levels - Failed: Image mismatch for example 3: {'rms': np.float64(0.08961256746282698), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference...
FAILED tests/integration/test_contour_plot_examples.py::test_example_4_north_pole_stereographic - Failed: Image mismatch for example 4: {'rms': np.float64(0.5533641260245302), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference-...
FAILED tests/integration/test_contour_plot_examples.py::test_example_5_south_pole_with_boundary - Failed: Image mismatch for example 5: {'rms': np.float64(0.5826523379800569), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference-...
FAILED tests/integration/test_contour_plot_examples.py::test_regression_npstere_da193_image - Failed: Image mismatch: {'rms': np.float64(0.011906453533254112), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference-example-imag...
FAILED tests/integration/test_contour_plot_examples.py::test_example_7_lat_pressure_zonal_mean - Failed: Image mismatch for example 7: {'rms': np.float64(0.18799208017491612), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference...
FAILED tests/integration/test_contour_plot_examples.py::test_example_8_log_scale_pressure - Failed: Image mismatch for example 8: {'rms': np.float64(0.11046718985438769), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference...
FAILED tests/integration/test_contour_plot_examples.py::test_example_19_multiple_subplots - Failed: Image mismatch for example 19: {'rms': np.float64(4.15330387349586), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference-e...
FAILED tests/integration/test_contour_plot_examples.py::test_example_21other_rgp_plasma - Failed: Image mismatch for example 21other: {'rms': np.float64(2.4241767024258962), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/refe...
FAILED tests/integration/test_contour_plot_examples.py::test_example_22_rgp_rotated_projection - Failed: Image mismatch for example 22: {'rms': np.float64(0.8955457536170187), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/reference...
FAILED tests/integration/test_contour_plot_examples.py::test_example_23_rotated_grid_axes_overlay - Failed: Image mismatch for example 23: {'rms': np.float64(0.28493929640571825), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_contour_plot_examples.py::test_example_34_lambert_conformal - Failed: Image mismatch for example 34: {'rms': np.float64(0.07193669150389208), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_contour_plot_examples.py::test_example_35_mollweide_projection - Failed: Image mismatch for example 35: {'rms': np.float64(0.21729055554358212), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_contour_plot_examples.py::test_example_36_mercator_projection - Failed: Image mismatch for example 36: {'rms': np.float64(0.22801506410925787), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_contour_plot_examples.py::test_example_37_orthographic_projection - Failed: Image mismatch for example 37: {'rms': np.float64(0.19407875641259678), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
FAILED tests/integration/test_contour_plot_examples.py::test_example_38_robinson_projection - Failed: Image mismatch for example 38: {'rms': np.float64(0.22239464014539573), 'expected': '/Users/bnl28/syncthing/Repositories/cf-plot/tests/referenc...
======================================================= 24 failed, 72 passed, 3 deselected in 46.75s =======================================================

Steps to reproduce

Attempt to run tests in a different environment than those in which it was built.

Your environment

The version of cf-plot and its dependencies in which you are encountering an issue.

Platform: macOS-15.7.7-arm64-arm-64bit
Python: 3.12.13
packaging: 26.0
numpy: 2.4.3
cfdm.core: 1.13.0.0
udunits2 library: /Users/bnl28/miniforge3/envs/work26/bin/../lib/libudunits2.dylib
HDF5 library: 1.14.6
netcdf library: 4.9.3
netCDF4: 1.7.3
h5netcdf: 1.8.1
h5py: 3.16.0
pyfive: 1.2.0.dev79+g34ae13ead
zarr: 3.1.5
fsspec: 2026.2.0
scipy: 1.17.1
dask: 2026.3.0
distributed: 2026.3.0
cftime: 1.6.5
cfunits: 3.3.7
cfdm: 1.13.0.0
esmpy/ESMF: not available
psutil: 7.2.2
matplotlib: 3.10.8
activestorage: 0.4.0
cfplot: 3.5.0
cf: 3.19.0

Expected behaviour

Tests should pass

Actual behaviour

Too many fail, but when you investigate they're all trivial:

As anb eaxmple: