I noticed that the thermodynamical data at `workflow.thermodynamics` as produced by `phonopy` does not match the data reported under `run.calculation.thermodynamics`. This is probably some kind of a small unit conversion issue. E.g. temperatures differ by a magnitude of 1000, and the data at `run.calculation.thermodynamics` seems to be the correct one. This conversion issue applies also for `heat_capacity_c_v` and `vibrational_free_energy_at_constant_volume`. Can you @alvin take a look at some point, or just tell me if you want me to apply the fix.