led to huge derivatives (and big, unrealistic changes) to the binary
orbit on binary_c timescales.

instead, the short disc timestep changes are integrated in time, and then
converted to average derivatives over the longer binary_c timestep.

added in the correction to the angular momentum flux Adam suggests.

recoded how mass changes are handled so all the units are converted from
the disc code's cgs to binary_c's units in the same place