possible bug in optimization step if the shape is a Lorentzian

[jobo322]

the default value of 2 is insufficient to sampling the necessary xy data to obtain a good optimization of FWHM parameter.

with a synthetic lorentzian shape the true FWHM is 0.15. The peak picking detect an approximation of 0.1572 for fwhm. But the optimization step returns the min limit of the search space for FWHM parameter

increasing the value of factorLimits to 15 - 20 increment the number of points in the decreasing shape of the lorentzian. It seems a scaling bug

[jobo322]

Another way to avoid this problem is to pass the baseline option, it is expected as a number.

[jobo322]

by refactor ml-spectra-fitting to rescale only the absolute max value to 1, the need to increment the factorLimit or pass a baseline value is not needed.