Supercell enumeration does not work as expected for 2D systems

Originally reported in #466 (closed) by @pernillatanner

generate_target_structure uses enumerate_supercells to create supercells that are multiples of the cluster_space-primitive. Since the cluster_space-primitive has to have periodic BCs in all directions even for 2D-systems, I end up with supercells that are stacked in z-direction (separated by vacuum). On that note, it would also be nice to have more control over the supercells than by max_size, e.g. I would like to be able to include 3x3x12 cells, but not 1x9x12. I currently use a workaround where I input a list of supercells to the generate_target_structure-function, which could be an optional argument.