relax SVN archive

Refactored frame_order.ave_pos_translate into the frame_order.average_position user function.

This is now used to set up all characteristics of the movement to the average domain position
including which pivot position to use for the rotation and if translation should be used.

Created 8 frame order system tests for the rigid test molecule.

Improvements to the frame order displacement test data script.

A relax log file has also been added from the running of the script.

Converted the tensor rotation in the frame order target functions to be a forwards rotation.

This allows the RDC to match the PCS, in that the average domain position is moved by a forwards
rotation and forwards translation.

Fixes for the PCS with translations and rotations.

The reverse rotations previously used have been converted to forwards rotations. This makes more
sense from the perspective of moving the original structures to the average position.

Elimination of all of the multi-processor code from the frame order target functions.

At no point did the parallelisation overcome the problem of huge amounts of data transfer, making
all multi-processor attempts run slower than the single-processor.

Import fixes and clean ups for many of the frame order auxiliary functions.

Fix for the frame order func_rigid() target function.

The Euler angles were being set to the translation values when translation is activated.

Fix for the calc_vectors() frame order target function method.

As the rotation occurs first, the centroid must not be displaced.

The atomic positions for the PCS are now rotated about a centroid for the frame order analysis.

This is for determining the average domain position. The centroid is currently set to the CoM.
Previously the rotation was about the pivot of the motion, but this does not make sense if the
starting structure is not in the correct position.