Fix cirrus ci and mpl (#4087)

* fix mpl SymLogNorm

* update graphics tests phashes

* add whatsnew entries

* gallery image phash update

docs/gallery_code/general/plot_anomaly_log_colouring.py

@@ -68,14 +68,14 @@ def main():
     # Create a 'logarithmic' data normalization.
     anom_norm = mcols.SymLogNorm(
         linthresh=minimum_log_level,
-        linscale=0,
+        linscale=1,
         vmin=-maximum_scale_level,
         vmax=maximum_scale_level,
     )
     # Setting "linthresh=minimum_log_level" makes its non-logarithmic
     # data range equal to our 'zero band'.
-    # Setting "linscale=0" maps the whole zero band to the middle colour value
-    # (i.e. 0.5), which is the neutral point of a "diverging" style colormap.
+    # Setting "linscale=1" maps the whole zero band to the middle colour value
+    # (i.e., 0.5), which is the neutral point of a "diverging" style colormap.
 
     # Create an Axes, specifying the map projection.
     plt.axes(projection=ccrs.LambertConformal())

docs/src/whatsnew/latest.rst

@@ -144,6 +144,12 @@ This document explains the changes made to Iris for this release
    developers to easily disable `cirrus-ci`_ tasks. See
    :ref:`skipping Cirrus-CI tasks`. (:pull:`4019`)
 
+#. `@bjlittle`_ and `@jamesp`_ addressed a regression in behaviour when using
+   `conda`_ 4.10.0 within `cirrus-ci`_. (:pull:`4084`)
+
+#. `@bjlittle`_ updated the perceptual imagehash graphical test support for
+   `matplotlib`_ 3.4.1. (:pull:`4087`)
+
 
 .. comment
     Whatsnew author names (@github name) in alphabetical order. Note that,