relative_height_cutoff clarification

[adas42]

This is thread to document the usage of the relative_height_cutoff parameter in the StrainAnalysisProcessor

[adas42]

I think relative height cutoff is working on the basis of the highest peak in all the spectra in a particular detector. This is surprising behaviour.

[rolfverberg]

The relative height cutoff is working as it should, but there is unexpected behavior when the baseline subtraction leaves a significant background in some low amplitude spectra:
possible solution: construct and subtract baseline on a per fit basis

BUG: If a fit is bad but accepted it can put the centers in the wrong initial spot to fit the next fit, even when the next fit has clearly defined and significant peaks.
solution: don't propagate the center, but initialize for each spectrum or do a scipy peak_find for each fit

[rolfverberg]

Use strain min/max instead fwhm_max/min, also for centers_range
DONE: added a centers_range_fraction input option to the detector config

[rolfverberg]

make the default for EDD to NOT use the previous map point as init condition.
DONE

[rolfverberg]

Added scipy find_peaks for each spectrum to get initial parameters for height, width and center and use the heights with rel_cutoff_height ot toss out small peaks before starting the fit.