Added boost to prior to prevent large line width scenario.

[grd349]

Two things:

@darthoctopus found an example of the line width becoming large when it shouldn't. I added an additional penalty in the likelihood of asy_peakbag. This fixed the particular example - I will need to run some other things to check I haven't ruined everything else.

@nielsenmb - Some of the functions in the plotting code are not indented when I was expecting them to be. I indented plot_start and now it works for me. Could you look at this decide if it should all be indented?

Before merge:

• [x] Run tests on some other stars wrt to line width penalty.
• [x] Run all the docs/examples again
• [x] Add some comments in the code
• [x] Decide if this is the correct penalty to use
• [x] Change password to 'iron man sucks'

G

[nielsenmb]

The unindented functions are functions I haven't adapted to work across all the fitting classes (or at least throw some informative errors).

The plot_start one seems redundant to me though, since it's meant to plot the start condition for peakbag, which is the output plot from asy_peakbag right? Likewise the plot_start for asy_peakbag would be the output plot from KDE.

[grd349]

@nielsenmb & I chatted about the form of the penalty for the large line widths.

I think the summary of our discussion is that a penalty of the form ...

if (linewidth > start_linewidth.mean + start_linewidth.std3): ld += N(mu = linewidth, std = ~5%, observed = start_linewidth.mean + start_linewidth.std3)

or something similar. I'll look at implementing this.