relax SVN archive

Missing import!

Added another check to iso_cone_S_to_theta() for bad order parameters.

This time the check is for order parameters less than -0.125.

Implemented Monte Carlo simulations for the N-state model.

Created a API common set_selected_sim() method for a single global model.

The method is called _set_selected_sim_global() and simply puts the select_sim bool list into the
cdp.

Created a common specific code model_loop() method for a single global model.

The method is stored as _model_loop_single_global().

Fix for bug #13259, the failure of model elimination in the full analysis script.

The problem was that model elimination is only performed on the current data pipe, so it should be
performed in the multi_model() method just after the optimisation of the individual model-free
models.

Fix for the test_grid_search_argfail_inc() unit test as the inc arg can be a list of None.

Objects placed into the relax data store structure are now stored in the XML save file.

They are also restored again with state.load().

Added a catch in iso_cone_S_to_theta() for bad order parameters > 1.

Now the function will return a cone angle of zero. Simplex optimisation will set S > 1.

Fixed some calls to maths_fns.order_parameters.iso_cone_S_to_theta().

The wrong function was being called by the free rotor isotropic cone model.