The egs++ angular spread source has unintuitive input

[rtownson]

To define the Gaussian by FWHM, it takes a negative sigma input. We might as well just add a fwhm input to make this more user friendly.

[ftessier]

Yes, I fully agree! Eventually, we ought to get rid of all cases where negative numbers are used as flags (yes, this includes negative energies for new histories in phase space files). But let's add fwhm right away! Now we do have to guard against negative "sigmas" anyways, so we might add a fatal error message that includes a note about this change of behaviour!