Added logarithmic anomaly contouring example.

[pp-mo]

Looks like another version of #1040, but actually rather more different than similar.

TODO:

• cross-refer to Added a custom colour plotting example. #1040 when that is merged.

[esc24]

I haven't looked at this in detail, but given the similarity to #1040 I suggest bringing some of the latest changes from that one into this PR.

[pp-mo]

I suggest bringing some of the latest changes from that one into this PR.

Done -- see pp-mo@889836e.

[pp-mo]

Now rebased to include merged code from #1040.
Each example now cross-references to the other.

[rhattersley]

It'd be nice to be consistent and symmetric with the use of the decimal point, i.e. one of:

• [-3.0, -1.0, -0.3, -0.1, 0.1, 0.3, 1.0, 3.0],
• [-3, -1, -0.3, -0.1, 0.1, 0.3, 1, 3],
• or perhaps even [-3, -1, -.3, -.1, .1, .3, 1, 3].

[rhattersley]

Thanks @pp-mo - I like it. 👍

Once we've discussed the two minor points I've raised above I'll be happy to merge.

[ajdawson]

This reads like the "similar cases" are logarithmic scaling, but you are referring to linear colour scaling aren't you? I don't think it reads well. You should be specific about this e.g.:

For linear colour scaling it is perfectly practical to select a suitable colormap and allow
contour to pick the level colours automatically from that, provided that the data range is
symmetric about zero, and that the colormap is diverging (see the 'Diverging colormaps'
section in: `<http://matplotlib.org/examples/color/colormaps_reference.html>`_).

Something along these lines?

[ajdawson]

@pp-mo - I've made several comments on writing style. For each I provided an alternative wording, please take this as an example of what I mean rather than an instruction to use it as a direct replacement for your wording.

[LukeC92]

@pp-mo Based on the comments, this pull request is similar to #1040; it's also obviously very old. Is there anything left from these commits that's worth adding to the current version of Iris? Might it be worth closing this request?

[pp-mo]

@LukeC92 thanks, on reviewing this I think I'm happy to close.
In retrospect, it seems too much about Matplotlib techniques, and not really Iris-specific.