Timeseries of matchups for various versions of harmonisation

[gerritholl]

In order to debug the post-harmonisation problems, plot timeseries of various versions of debug FCDR radiances, for the various fields added in #331 (due to #272). To get there:

• The matchup combiner should be able to combine from multiple versions, rather than just one,
• The enhanced harmonisation files should output lots of additional "nominal_measurand" values and perhaps also additional measurement equation inputs,
• The harmonisation input file analysis scripts should then not only plot overall statistics of those but also timeseries.

See also #206.

[gerritholl]

Complication: the matchup combiner works on an unharmonised version, at least to generate the harmonisation inputs, but for these timeseries we are interested in post-harmonisation.

[gerritholl]

But does that actually matter? The harmonisation output files are primarily writing out inputs, not radiances/BTs, so it can work as a post-harmonisation analysis too — part of needing to iterate the Ks anyway.

[gerritholl]

This should not be various versions, the relevant data are all in the v0.8pre2b debug version.