The question came in mind is why you need it in parallel ? Boltztrap just process DFT data and run no-costly calculation relatively to DFT. that's why it is not designed to run in parallel mode I think.
it never exceed one hour for a single core for me !
first check your calculation ie scf is parallel or not, only then something useful could be suggested....nothing more is required to make to run boltz parallel, internally it works in single core mode by default