relax SVN archive

For the frame order displacement test suite data, an optimisation script has been added.

This is to verify this data through the auto-analysis, but it is not part of the test suite as these
tests are insanely slow (~8 hours on one machine).

Large decrease in the time required for the pseudo-ellipse CaM frame order system test.

The number of integration points has been massively decreased from 1000 to 15 (all other tests are
50 points). This halves the time down to about half a minute on one machine.

Decreased the total time required for the frame order system tests.

Instead of using all the PCS data, the PCS data subsets are now used. This decreases the total time
on one machine from 656.300 to 461.354 (total CPU seconds from profiling).

Fixes for the chi2 values in the CaM frame order system tests.

As forwards rotations are now used for the average domain position, the low precision of the
quasi-random integration means that the chi2 value is slightly different.

Converted all of the CaM frame order system test scripts for the new average position notation.

The reverse angles have been converted to forwards angles.

Fix for the rigid frame order system test script for the frame_order.average_position user function.

The old frame_order.ave_pos_translation user function was being used.

Removed all references to 'Monte Carlo integration' from the frame order target functions.

This was replaced by the quasi-random numerical integration a long time ago.

Removed the remaining residual, slow frame order multi-processor code from the target functions.

The CaM frame order system tests now use the obligatory frame_order.average_position user function.

The grid search for the frame order auto-analysis with translation is now handled differently.

This is to avoid getting trapped into a local minimum. The grid for this split into a grid search
over the rotational space first, then a grid search over the translation. This is repeated for good
measure. This is much more accurate and reliable for finding the global minimum and is also much
faster.