[newchem-cpp] Hoist mask-modifications and edot initialization out of cool1d_multi_g

[mabruzzo]

To be reviewed after #545

---

This PR is pretty straight-forward.

The main changes:

• we now modify itmask (based on the temperature floor) before we enter cool1d_multi_g
• we now fill itmask_metal before we enter cool1d_multi_g
• we now initialize edot (i.e. fill with zeros) before we enter cool1d_multi_g

The last change is probably the most controversial change. This change was motivated by the fact that I want to be able to compute all cooling and reaction rates pertaining to our dust model all at once (so we can better compartmentalize that code and significantly improve performance/reduce allocations for Gen's model). At present, the easiest way to handle this will be to call that logic before we call cool1d_multi_g (i.e. when we call cool1d_multi_g, the edot buffer will already include cooling contributions from the dust model).¹

This PR and the prior PR (as well as the one before that) should all use the same gold-standard.

Footnotes

1. If we want to call cool1d_multi_g before the dust calculations, then we're going to need to move the optical-depth adjustments to a separate function called after the dust calculations. If there's a preference for this plan, we can totally do it -- I don't really which care which approach we take. ↩

[brittonsmith]

I like the extraction of the masking to external functionality, but it might be worth stuffing it into an initialize_masks function so we're not duplicating so much.

I have noted two additional places where metal masking is happening in an identical without the use of itmask_metal. I think we should thread this through there. These are:

• calc_tdust_3d.cpp
• make_consistent.cpp
  On a similar note, I also note that calc_tdust_3d.cpp is still doing its own metallicity calculation.

I've now written and rewritten thoughts about extracting the dust component to the edot calculation three times. I was initially not in favor as I was concerned about code duplication to get cooling time, but I'm trying to keep in mind your point about making it easier to modify dust treatments, so I think it makes sense. In any case, I'm willing to explore it and see how the initial refactor plays out.

Update: in light of moving on to PR #547, you can definitely consider the last point above as requiring no action, but let me know your thoughts on the first two points.