relax SVN archive

Added the RMSD determined via showApod for the 69 experiments.

Work in progress for Support Request #3083, (https://gna.org/support/index.php?3083) - Addition of Data-set for R1rho analysis.

Updated the Relax_disp.test_ns_r1rho_3site_linear system test so it now passes.

The chi-squared value is not exactly zero as there are numerical differences between relax and
cpmg_fit due to different approaches being used.

Updated the relax results for the Fyn SH3 R1rho dispersion data.

This is for the recent changes to the 'NS R1rho 2-site' dispersion model.

Modified the 'NS R1rho 2-site' dispersion model to match the 'NS R1rho 3-site' models.

The 6D evolution matrix indices have been rearranged to match the 9D matrix indices. The tilt angle
for the initial magnetisation is no longer that for the average offset but that of state A, as was
changed for the 'NS R1rho 3-site' models earlier (r21911). The system test was therefore updated
for the slightly different behaviour.

Updated the relax results for the synthetic data of the 'NS R1rho 3-site linear' dispersion model.

Added the Omega_A,B,C resonance offset parameter definitions to the dispersion chapter of the manual.

Docstring fix for the lib.dispersion.ns_r1rho_3site module.

Fixes for swapped indices in the relaxation evolution matrix for the 'NS R1rho 3-site' dispersion models.

Last fixes for the 'NS R1rho 3-site' dispersion models.

These now behave identically to the cpmg_fit program with the PEAK_SHIFT flag disabled. The tilt
angle for the initial magnetisation is no longer that for the average offset but that of state A.

Transposed some of the 'NS R1rho 3-site' model evolution matrix elements.

These now match the 'NS R1rho 2-site' model.