This may need to be split into two issues: there is the standalone program, but also integration with at least Gaussian and Q-Chem. If we could invoke a parser inside another parser (also desirable for #808) then a lot of duplicate work would be avoided.