Goodness of fit for varying parameters

[rmartinho]

Given #36, it is only natural to automate verification of fitness to different models, in order to estimate the likelihood of some code belonging to a particular complexity class.

[arximboldi]

I just realized that #56 partially addresses this, because Plotly already supports fitting data over the summary plot. To get a fit, one can follow these steps:

1. Generate a plot using the HTML reporter. For example: https://sinusoid.es/misc/nonius/params-example6-log.html
2. Click the "Save" (floppy disk icon) button on the controls that appear over the plot. This brings the plot to the Plotly platform, allowing further editing.
3. Click "Fit data", which opens a panel to choose various fit functions and presents the results with their errors, etc. The resulting fit is also shown in the diagram.

[rmartinho]

That's a pretty nice feature!

I'd rather if nonius had this built-in for convenience, though, otherwise, CSV output would be enough ;P I know people who basically just do CSV output and then do all stats in MatLab.

I'll keep it open for now.