Peak fitting algorithms

[creuzige]

I wanted to get some feedback/discussion about the new peak fit functions Max added (as of commit 1b8160a.

1. For the move left and move right functions, I thought it may be good to just recycle the fit_peaks function, rather than having a secondary fitting implementation. Thoughts?
2. It looks like the move left and move right resets the entire peak list, not just the peaks that failed to fit. Would we be better off just focusing on the peaks that didn't work? It looks like it's possible to set which peaks are refined (peaklist parameter in set_peakFlags, see GSAS II documentation
3. It looks like the first peak fitting cycle (where we're just varying intensities) is where the negative intensities first show up and even when position is varied next the program seems locked onto these values. It may be worth just doing that step with a few different positions to try and lock onto a peak. Thoughts?

I know sometimes research analysis code seems like building the plane as we are flying it :)

[creuzige]

Checked, you can use LeBail in Scripting toolkit. Maybe try that or to a fit to find the peak locations with a common lattice.