Chiral twin unit test

[mducle]

As part of unit testing of the spinw.addtwin method (PR #85) a restriction was imposed on the form of the transformation matrix specifying the orientation of a twin - previously it could be any 3x3 matrix, but now it must be a valid rotation or reflection.

There was some discussion at the time as to whether this is too restrictive - specifically, whether in the case of a chiral system whether this is restriction is obeyed or not.

Searching through the literature, I've only been able to find a few examples of a chiral twinned system (e.g. CuFeO2), and I've not been able to find any which violates the restriction we imposed. (In the case of CuFeO2, there are 3 twin domains related by a 120 degree rotation, as the system has hexagonal symmetry).

This issue is here as a reminder that the restriction may be violated and to revisit this in future.

[mducle]

[After discussions with D. Khalyavin]: The number of domains and transformations between them is determined by the symmetry change between the parent and distorted structures (for crystal structure changes) or between the paramagnetic and ordered structures (for magnetic structures). The actual allowed transformations can be determined by group theory, using a program like ISODISTORT and includes point operations on the lattice (rotation, inversion, mirror) which may be combined with translations of the origin. I'm not sure how the translation affects things - especially since we are dealing with a transformation of reciprocal lattice coordinates whilst the symmetry operators from group theory applies to real space...

@RichardWaiteSTFC In any case, I think we should add the inversion (-eye(3)) operator as an allowed twin transformation operation in addition to rotations and reflections.