Don't use a magic `0.6` in `proc_base.gm` about nmrglue

[jjhelmus]

I like the idea of adding a note that points out that the 0.6 comes from 1.0 / (2 * sqrt(log(2))) in the docstring of nmrglue.proc_base.gm but I do not like the idea of changing the value in the function itself. For better of worse, the most "standard" definition of the apodization functions the NMR community has are those in NMRPipe and I think nmrglue should use these. Also the gm function is named for NMRPipe's GM function, so I think keeping the results as identical as possible would be best. What I would be in favor of is adding a gauss or gaussian apodization function which applies an exp(-g*_2) based window, using the 1.0 / (2_sqrt(log(2))) constant.

A good explanation of why 1.0 / (2*sqrt(log(2))) * pi * g is used in the definition of the Gaussian apodization is this using this definition the apodization results in the addition of g units of Gaussian line broadening or, put another way, this transforms into a Gaussian function with FWHM of g. Working along these lines would make a good addition to the docstring. The math behind this is:

FFT[exp(-(a * pi * g * t)**2)] = exp( -(pi * f / (a * pi * g))**2) = exp(-(f / (a_g))_*2)

The full width half maximum of this Gaussian is 2 * a * g * sqrt(ln(2))

If a is 1/(2*sqrt(ln(2)) then the FWHM is g.