relax SVN archive

Started to make changes for multiple field relaxation disperison analysis.

This seems necessary, so maybe we should not support single field analysis of relaxation dispersion
at all.

-> Kovrigin et al. (2006) JMagRes, 180: 93-104.

...

The changes made here are only a first draft and may not work. In particular, maybe the
spectrum.read_intensities(), relax_disp.cpmg_frq(), spectrum.replicated(),
spectrum.error_analysis(), and deselect.read() functions will need to know the magnetic field to
which the particular dataset is associated...

In fact, the different datasets should be input first and their R2eff calculated indepentantly. In
a second step, the actual relaxation dispersion curve fitting should be made with all data.

Fixed a bug where the 'id' argument was not set.

This was proposed by Ed in a post at:
https://mail.gna.org/public/relax-devel/2009-01/msg00127.html
(Message-id: <7f080ed10901280059y3303254bv89c0c1293e9916c0@...>)

The members list is now stored in the pseudo-spin to map this to the respective spin containers.

One more fix for test_create_pseudo_spin().

Fixes for test_create_pseudo_spin() and test_create_pseudo_spin2().

The pseudo-spin name and number is now stored in the member spin containers.

This gives a mapping between from the spin to the pseudo-spin.

Modified the checks for the pseudo-spin creation unit tests.

Wrote a second unit test for the creation of a pseudo-spin.

Added some more checks for the test_create_pseudo_spin() unit test.

Changed the default value for 'int_cpmg' to avoid an impossible mathematical situation: ln(0).