relax SVN archive

Created the Structure.test_read_gaussian_strychnine system test.

This will be used for implementing and testing the structure.read_gaussian user function.

Added another Gaussian log file of strchnine, this time with DFT structure optimisation.

The file is bzip2 compressed to save space.

Added the 'NS MMQ 3-site' models to the dispersion variables.

This is for the 'NS MMQ 3-site' and 'NS MMQ 3-site (linear)' CPMG dispersion models.

This follows the tutorial for adding relaxation dispersion models at:
http://wiki.nmr-relax.com/Tutorial_for_adding_relaxation_dispersion_models_to_relax#Adding_the_model_to_the_list

Created the Relax_disp.test_ns_mmq_3site_branched system test.

This is for the 'NS MMQ 3-site (branched)' CPMG dispersion model.

This follows the tutorial for adding relaxation dispersion models at:
http://wiki.nmr-relax.com/Tutorial_for_adding_relaxation_dispersion_models_to_relax#The_test_suite

Created a relax script for back calculating R2eff values for the same parameters as cpmg_fit.

This is for the 'NS MMQ 3-site (branched)' CPMG dispersion model.

This follows the tutorial for adding relaxation dispersion models at:
http://wiki.nmr-relax.com/Tutorial_for_adding_relaxation_dispersion_models_to_relax#The_test_suite

A big change to the usage of the relax_disp.r2eff_read_spin user function.

Now the nu_CPMG frequency or the spin-lock field strength must be set prior to calling this user
function. This allows for more flexibility as often the experiment IDs and frequency values in the
files do not match to the same number of decimal places. The frequency is no longer read from the
file but must be preset.

The relax_disp.r2eff_read_spin user function now really strips comments and empty lines from the file.

Missing import of a RelaxWarning.

Created data for a 'NS MMQ 3-site (branched)' model using cpmg_fit from Dmitry Korzhnev.

Another fix for bug #21309, the 'R2eff' dispersion model failure when peak intensity data is missing.

The bug is reported at https://gna.org/bugs/?21309.

This second problem is only for the numeric CPMG models for when all data at one magnetic field
strength is missing. When the relaxation dispersion target function is being set up, the creation
of the self.power data structure holding the number of CPMG blocks fails. The problem is that the
relaxation time for the missing field strength is set to NaN. This is now caught using
lib.float.isNaN().