Unc calculation in frequency domain

[lucaob]

Added a method called FdData.calculateFDunc that uses the equations from Withawat paper.
Comparison results with the previous method FdData.calculateSTDunc results in:

• unc in abs, real and imag part of FFT are close with the two methods and very low but STDunc gives almost constand results while FDunc results are slightly more oscllating behavior
• unc in abs is completely different: STDunc is around 1.5 while FDunc is between 0.001 and 0.01 (measured data from Marburg measurements of INRIM Silicon sample with nitrogen)
• STDunc for imag part is zero!

Maybe some problems in the code.

FDunc on the phase

unc on the modulus (red line is FDunc)

unc on the real part (red line is FDunc)

unc on the imag part (red line is FDunc)

[DavidJahn86]

I dont know, but in frequency domain, it takes also very long to evaluate the uncertainty, changed the method slightly but still takes a long time (most of the time, when running terapy) Maybe we make uncertainty calculation optional...