Hi kwerner,
The new Goddard 4-ice scheme (mp_physics=7) calculates the hail contribution in subroutine saticel_s. And original paper (Bae et al. 2018) also shows the reflectivity from hail.
Currently, those microphysics schemes which support the option do_radar_ref are most 3ice schemes. Some schemes can let you choose graupel or hail but use the same matrix (ex: qg) to store them. That's why you can't see qh doesn't be passed in.
Saitama