Parsing error relativistic data
Under Method/Simulation/DFT/Relativity method, scalar_relativistic captures two different methods from CASTEP (Koelling-Harmon) and FHI-aims (zora scalar) under the same name. The latter should correspond to ZORA, not be confused with atomic ZORA, which is also implemented in FHI-aims [1].
All of these methods are scalar relativistic, so I suggest introducing Koelling-Harmon as its own category.