k-packet changes for free-free cooling

[jhmatthews]

These are changes to deal with errors in how I'd coded the free-free cooling channel in kpkt(). The main changes are:

• a fix for matrix ionization calculation overwrites macro_pops values #498 so that macro_pops ion populations are not overwritten when they shouldn't be
• fixes to do with macro atoms - one free free errors #492 - kpkt() always calls one_ff with a wide frequency range, from 0 to (100 * k T / h).
  if the photon is less than the global minimum frequency, we discard the photon with an istate set to P_LOFREQ_FF, in the same way we do for adiabatic cooling.
• some streamlining of istat code
• some extra reporting of cooling rates

IMPORTANT: There is still one outstanding caveat here, which is that I have to set kpkt_rates_known to -1 (i.e. not known) when restarting a run in the spectral cycles to avoid errors to do with kpkt destruction rates. It's not clear why this is, because once the model is converged the kpkt rates should be the same. It suggests a bug to me and should be investigated or discussed.

I'm opening the pull request now because this should be merged before southampton.