Change routine for y-space fit

[GuiMacielPereira]

Current routine has a few shortcomings:

• Option to symmetrise the data is not ideal, and should be replaced by subtracting FSE to the first NCP peak
• Errors on the parameters, especially for functions like GC_C4, GC_C6 and GC_C4_C6 are often on the same magnitude of the parameter values, which invalidates the fit
• Global fit on all detectors is slow
• Positivity constraint is not being successfully implemented, since some fitted functions have negative values

Possible solutions and avenues to explore:

• Implement FSE corrections before focusing data
• Instead of convolving NCP to fit, deconvolve the data and fit NCP
• Explore using Mantid algorithm for plotting the error profiles of the parameters, maybe an alternative to using MINOS?
• Use Mantid Fit for global fit instead of using iminuit
• Use analytical solutions for the positivity constraints where possible, resort to brute force numerical range constraint for GC_C4_C6