Ruth Comer

Noting that the calendar previously called "gregorian" in the cf-conventions is now called "standard", I've updated the comments through this issue to hopefully reduce confusion.

I have stuck up a PR based on my comments above at #5138

That PP load&extract gets a lot quicker if you simplify the latitude constraint ```python lat_point = c.coord('latitude').cell(45) lat_const = iris.Constraint(latitude=lat_point.point) ``` Possibly because of https://github.com/SciTools/iris/blob/caae45075795c1e13fbb0f04488c3b99f15b60e8/lib/iris/_constraints.py#L308-L315

Maybe it could run specifically in PRs that change `imagerepo.json`, which are not all that frequent.

If we are thinking of hashing all the reference images as part of the regular tests, could we ditch the `imagerepo.json` entirely and just have the individual image tests compare...

Do we know why repo review advises the code to be under `src`? A quick check of the big four scientific python packages show three of them (numpy, pandas, scipy)...

Matplotlib has two versions of `inset_axes`. You appear to be using the [axes_grid version](https://matplotlib.org/stable/api/_as_gen/mpl_toolkits.axes_grid1.inset_locator.inset_axes.html#mpl_toolkits.axes_grid1.inset_locator.inset_axes). There is also an [axes method version](https://matplotlib.org/stable/api/_as_gen/matplotlib.axes.Axes.inset_axes.html#matplotlib.axes.Axes.inset_axes). `indicate_axes_zoom` is designed to work with `Axes.inset_axes` whereas the...

> I just tried to remove the the transform from the indicate_inset_zoom and I get this hybrid behaviour. I think in the first plot matplotlib is trying to connect some...

With a few tweaks in Matplotlib, we can have more interesting combinations https://github.com/matplotlib/matplotlib/pull/29174

@that-ar-guy please do!