Make ti-cool selectable at runtime

[mabruzzo]

This PR builds on #415

---

Essentially, this makes the TI cooling curve select-able at run-time. To select TI-cool, use should pass piecewise-ti to the chemistry.kind parameter (which was introduced in #415)

This PR involved some incidental refactoring:

• I replaced the TI_cool function with the CoolRecipeTI class. The functionality of TI_cool is now implemented by the CoolRecipeTI::rate_cool member-function. This was done to match how other cooling recipes are implemented
• I slightly refactored how we implement cooling for T>1e4 K for CIE cooling and TI cooling. Previously, but regions of code used the same piecewise analytic formula, but duplicated the implementation of it. Now we just implement it once.
• I refactored how we implement photoelectric heating gets implemented.
  • previously, photoelectric heating was implemented by a function Photoelectric_Heating and cloudy-cooling separately tracked the n_av_cgs parameter. If we made the CoolRecipeTI class mirror the architecture of the cloudy recipe, the class would also need to track n_av_cgs.
  • This is a little undesirable because n_av_cgs is an implementation detail of a particular photoelectric heating model. Robustly documenting the code's limitations would involve repeating some warning-comments. Plus it would be a little messy to replace the photoelectric heating model.
  • Now, we encapsulate the photoelectric heating term in a PhotoelectricHeatingModel callable.
  • (in the future, it might be better to not store photoelectric heating separately from the rest of the cooling recipe)

While I was here, I slightly streamlined the implementation of CoolingUpdateExecutor (I removed an unnecessary indirection that was only in place to simplify the number of changes in #415). I also added an overview of how the code is supposed to work to the top of the file.

Important

It's worth noting that the lower Temperature sections of TI-cool cooling function comes from papers where they assumed $\mu=1.25$ and the higher Temperature is for temperatures where we can safely assume $\mu=0.6$.

• Currently, we assume $\mu=0.6$
• We could conceivably do some kind of special renormalization to account for changes in $\mu$, but I'm a little hesitant to go down that route