Slab orthogonaliser routine breaking ISHIFT = 4

[nedtaylor]

The issue occurs for non-tetragonal cells:

• Atoms that should be at 0.0 along c are given values of 1.0 (computational rounding errors)
• A vacuum is placed between the atoms at 0.99 and 1.0
• The orthogonaliser routine is applied to the slab, shifting the atoms at 1.0 differently from the atoms at 0.0
• Then, when the cell is then renormalised, these atoms are renormalised to 0.0 along c, but are now are then incorrectly shifted (by the orthogonalisation amount) with respect to the rest of the slab.