relax SVN archive

A few small fixes for the generate_data.py script.

The bond length is now 1.02 Angstrom.

Removed the grid search and started optimisation of the ellipsoid diff tensor at the solution.

Fix for the internal structural object PDB writing.

The PDB format was not being followed correctly! The TER record comes after ATOM records but before
HETATM records. It is not needed if there are no ATOM records.

Small fix for generate_vector_dist() for when no atoms currently exist in the molecule container.

The diffusion tensor optimisation script now optimises.

Added errors to the diffusion tensor relaxation data.

Added the infrastructure to start to check the diffusion tensor optimisation.

The relaxation data from diffusion tensor script now calculates the data at multiple field strengths.

The new data has also been added to the repository.

Added relaxation data for a diffusion tensor with no internal motion.

A script has been added to generate a uniform distribution of vectors (output to PDB) and the
corresponding relaxation data for a simple ellipsoidal diffusion tensor. All vectors are rigidly
attached to the molecule.

Final bug fix for the ellipsoid() function for the (Dxx, Dyy, Dzz, Dxy, Dxz, Dyz) parameter set.

The 2304 unique Euler angle values for a rank-2 tensor with no skew (12 axis orders, 2 frames of
ref., 2 directions (forwards + reverse), 4 rotational symmetries (pi rotations about each
eigenvector), 2 handednesses (left + right), and 6 eigenvalue orders) are now all properly handled.
The resultant Euler angle values are a single set of the 2304, matching that already used in relax
(zyz, rotated axis system, reverse direction, rotational symmetry collapse to 0 <= alpha, beta,
gamma <= pi, right handedness, and Dx <= Dy <= Dz).