relax SVN archive

Reverted r11913 as this was completely wrong and broke the reading of results files.

The command used was:
svn merge -r11913:11912 .

Added a check to make sure the relax save states can not be loaded by results.read().

Added a system test to check that relax save files cannot be loaded as results files.

Updated some test data for the unit tests.

Last fix for bug #17361 (https://gna.org/bugs/index.php?17361) the loading 1.2.x sim data.

The frequency information is now correctly updated when data is missing.

Completed the test_read_results_1_2_pse() model-free system test for reading relax 1.2.x results.

Bug fix for the relaxation data add_data_to_spin() function - missing data was not properly eliminated.

For the 1.2 results files, the model information is now being setup even if None.

Another fix for bug #17361 (https://gna.org/bugs/index.php?17361) the loading 1.2.x sim data.

This time the global structures were too large - the lists were being appended to for each spin and
simulation, rather than only each simulation.

Fix for bug #17361 (https://gna.org/bugs/index.php?17361).

The chi2_sim structure should have been initialised to a list, not a dictionary.