[newchem-cpp] Consistency tweaks: GrainSpeciesCollection and cool_multi_time_g

[mabruzzo]

This PR was originally proposed as brittonsmith#25

---

To be reviewed after #331 is merged (i.e. the PR where solve_rate_cool_g was transcribed)

---

This PR essentially goes back and makes some tweaks to the internal GrainSpeciesCollection helper-type and the C++ version of the cool_multi_time_g function, both introduced in PR #275, to achieve consistency with some of choices made in #331¹

This is actually a fairly small PR. The changes include:

1. Switching the GrainSpeciesCollection to be more consistent with the similar, but more generic SpeciesCollection type. For concreteness:
   • when I created the SpeciesCollection type, I designed it to use enumerators defined within the SpLUT pseudo-namespace as a compile-time lookup table (mapping species names to an index)
   • this PR's 1st commit refactors GrainSpeciesCollection so that it uses the the enumerators within the newly-defined OnlyGrainSpLUT pseudo-namespace as a lookup-table.²
2. Various superficial tweaks to the cool_multi_time_g function:
   • it now uses the InternalGrUnits type instead of manually tracking the utem, urho & uxyz floating point values and an instance of the code_units struct (this also removes some boiler-plate from local_calculate_temperature)
   • modified the j,k loop to use grackle_index_helper and IndexRange (the latter helps removes some ambiguities about 0-based and 1-based indexing when calling routines that are not yet transcribed)
   • thanks to the previous two changes in the last 2 bullets, I was able to replace all FORTRAN_NAME({subroutine}) occurrences with the nicer, C++ wrapper functions. Recall that these wrappers take shorter argument lists, mostly composed of structs, and internally forward the struct-members on to the underlying routine. This is important for the future process of transcribing a routine like cool1d_multi_g, which is called from multiple other transcribed routines.³
   • I also made some superficial tweaks to the 2 for-loops within cool_multi_time_g. The variables holding variable values now only hold the 0-based index values (rather than 1-based index values that we adjust every time that they are used). This was not strictly essential to this PR
   • we now pass 2 arguments by value instead of by pointer (for consistency with solve_rate_cool_g). This was not strictly essential to this PR

Footnotes

1. To put it another way, PR [newchem-cpp] initial transcription of cool_multi_time_g #275 was my first attempt at transcribing a top-level function. PR [newchem-cpp] transcribe: solve_rate_cool_g to C++ #331 was my second attempt at doing so. I learned a lot while working on [newchem-cpp] transcribe: solve_rate_cool_g to C++ #331 and established some conventions that improved quality and made transcription of certain aspects of the code a lot easier. Now I'm going back and applying this experience to improve the quality of the code written as part of my first transcription attempt. ↩

2. In the future, this change is going to significantly simplify the C++ versions of calc_all_tdust_gasgr_1d_g and calc_grain_size_increment_1d. Currently these routines have multiple instances where the same exact logic is repeated 12 times (once per dust grain species). This change should help us replace instances of this code-duplication with for-loops. ↩

3. Essentially, we want there to be just 1 location in the entire codebase where we directly reference any given Fortran routine before we transcribe that routine. (there are a couple of reasons for this that I could go into if you want) ↩

[mabruzzo]

Here is an example of "cleaned up indexing" (for the discussion in #275)

[mabruzzo]

Here is another example of "cleaned up indexing" (for the discussion in #275)