relax SVN archive

Added the insignificance dispersion auto-analysis argument to the Hansen CPMG data optimisation script.

Created an INSIGNIFICANCE variable for the relaxation dispersion sample scripts.

This is to allow the user to eliminate insignificant models.

The relaxation dispersion auto-analysis now accepts the 'insignificance' argument.

This is then used in the relax_disp.insignificance user function prior to the optimisation of each
model, so that spins with insignificant dispersion curves are not optimised. The 'R2eff' and
'No Rex' models are skipped for obvious reasons.

Improvements for the relax_disp.insignificance user function.

Text is now printed out when a spin is deselected. And all spins set to the 'R2eff' model are
skipped.

Created the new relax_disp.insignificance user function.

This will be used to deselect all spins whereby the maximum difference in all its dispersion curves
is below a certain cutoff.

Fixes for all of the system tests using Flemming Hansen's CPMG data subset.

The errors are now different and the new residue 4 has to be deselected and ignored.

Added all of the new NESSY plots for the truncated Hansen CPMG data.

Updated the software comparison document for a subset of Flemming Hansen's CPMG data.

This now includes residue 4, the changes in results for all software, new NESSY results due to fixes
I made in NESSY, and the new results for numeric model in ShereKhan.

Updated the ShereKhan results to include residue :4 and the ShereKhan numeric results.

The numeric model in ShereKhan was previously buggy and did not return results. This has been fixed
after I sent feedback to the authors.

Updated the NESSY log for the bug fix of r1105 (in the NESSY repository).