matrix ionization calculation overwrites macro_pops values

[jhmatthews]

This is related to other issues, e.g. #493 - in trying to address the issues with macro-atoms, I think I've found a bug in the matrix ionization code when macro-atoms are used.

What I thought was supposed to happen is that, inside the loop over electron density, the ionization rate matrix for simple ions is calculated followed by a call to macro_pops to get the macro-atom level populations and ionization states. However, it seems to me that the densities that are copied over to the newden array and these overwrite the macro-atom ones:

    for (nn = 0; nn < nions; nn++)
    {
      newden[nn] = 0.0;         // initialise the array element

      /* if the ion is being treated by macro_pops then use the populations just computed */
      if ((ion[nn].macro_info == 1) && (geo.macro_simple == 0) && (geo.macro_ioniz_mode == 1))
      {
        newden[nn] = xplasma->density[nn] / elem_dens[ion[nn].z];
      }

      for (mm = 0; mm < nrows; mm++)    // inner loop over the elements of the population array
      {
        if (xion[mm] == nn)     // if this element contains the population of the ion is question
        {
          newden[nn] = populations[mm]; // get the population
        }
      }


      if (newden[nn] < DENSITY_MIN)     // this wil also capture the case where population doesnt have a value for this ion
        newden[nn] = DENSITY_MIN;
    }
    free (populations);

Sorry to raise another issue, but this one is potentially pretty important. Hopefully I've misunderstood something here.

[jhmatthews]

Note that this doesn't affect in published work, but it does affect your TDE runs @saultyevil - it's not exactly "wrong" but it's not how we meant to do things and is inconsistent with the way we do stuff in the old non-matrix ion modes.