gaussianfit.py needs improvement

[pavoljuhas]

The gaussianfit.py extension e13ed35 is currently not very usable, for example the gaussianfit function does not return any result. I suggest the following improvements:

• replace header with CMI-related one as in diffpy/diffpy.srreal@302745e
• convert gaussianfit function to a class that would accept the same arguments
• the class should automatically pick meaningful starting values for x0, sig from the maximum and FWHM in the y-array
• the class should have attributes A, sig, x0, recipe, results, x, y, dy, yg, where
  • A, sig, x0 are properties forwarded to recipe.NAME
  • x, y, dy are read-only input data, yg the Gauss curve evaluated for the current A, sig, x0
• the class should have methods:
  • plot - replacement for plotResults()
  • refine() - call scipy leastsq for the current recipe state
• remove the "gau.dat" file and the if name == "main" block.
• write a few-line example in the class docstring that generates Gauss curve with some noise and executes the gaussfit class.

[pavoljuhas]

Kevin volunteered to fix this up.