relax SVN archive

Fixes for the print out of eliminated Monte Carlo simulations.

The previous print out was saying that the spin system is eliminated, which is incorrect.

Added a final 10 step optimisation of the 'sphere' global model.

This completes the system test.

Restructured the introductory docstring and added the references for all techniques in the script.

Fix for bug #12412 (https://gna.org/bugs/index.php?12412).

Caught the situation of when the global chi-squared value is missing when performing global model
model selection. This now throws a RelaxError explaining the situation.

Fix for the final model selection, the data pipes 'local_tm' and 'sphere' should be used.

Renamed a number of the data pipes to better reflect what they are.

Model elimination is no longer performed on the global model in the omp_model_free system test.

Fix for bug #12411 (https://gna.org/bugs/?12411).

The problem was with the duplicate_data() skipping the duplication of the spin specific data if the
new data pipe (which should start off as empty) contained no sequence data. Hence the data was not
being copied.

Bug fix for the duplication of sequence data during model selection.

The sequence.copy() function was not preserving the selection flag so that new option is sent into
the copy() and generate() functions, originating from the duplicate_data() model-free method.

Bug fix for the fix() function.

The fix flag for the spins was not being set when the 'all' element was sent into the function.