Make sure refinement results can be plotted

[MBartkowiakSTFC]

Description of work
Simplifies the plotting mechanism for the parameter output. The plotter takes all the points sampled by CMAES and makes a corner plot showing the parameter correlation:

There is also a scatter plot of parameters sorted by FoM:

It would probably make sense to output the correlation information directly from the optimiser. Also, discarding a fraction of the initial values before creating the corner plot could also help illustrate what correlations have been found.

Fixes

1. Remove GPO from PlotResults,
2. Add a scatter plot of parameters vs. FoM.
3. Include plot output in examples/sto-dos.py.

To test
Run examples/sto-dos.py. It should save plots to a file. (You can adjust the parameters to make the run faster/shorter.)
Add control.plot_results() to the "Argon-a-to-z" Jupyter notebook and check if the plot is created at the end.

[MBartkowiakSTFC]

@ChiCheng45 @oerc0122 What kind of visual output do you think would be useful at the end of an optimisation run? I was considering adding a plot of FoM over time to give a basic idea of how the refinement was converging.

[ChiCheng45]

This seems like something that someone would create themselves to suit their own needs/research project. But anyway, the above looks good to me and I agree some plot over time would probably be useful. Here is what I was creating for Argon.

I was also plotting the Log10(FoM).

[MBartkowiakSTFC]

OK, I added a logscale plot of the FoM values as a function of refinement step.

Also, the cornerplot inputs and the parameter plot inputs now both use the same subset of all the points, which currently means that they get sorted by the FoM and points with the highest FoM are not included in the plot. The default setting is to keep half of the points.