relax SVN archive

The fixed alignment tensors are no longer included in the parameter vector from _assemble_param_vector().

Added the N-state model _minimise_setup_fixed_tensors() method.

This assembles the fixed tensors into an array for use in the maths_fns code.

The align_tensor.copy() tensor_to arg can now be None.

This is useful for copying between data pipes.

Created the align_tensor.fix() user function for fixing the alignment tensor during optimisation.

The alignment tensor optimisation is now optional within the population N-state model target fn.

Fix for the Grace write_xy_data() function for when data values of None are passed in.

Fix for the get_data() function - the returned data structure is dependent on the errors.

Instead of a list of x, y, error 1, error 2 with the errors set to None if non-existent, only the
required errors are returned.

Fix for the unit tests, the pseudo_ellipse module is now listed in maths_fns.__all__.

Fix for the relax_io.read_spin_data() function for when None is encountered.

The None values are now returned rather than skipped so that the calling code can decide what to do
with it.

The RDCs and PCSs can now be given optimisation weights.