[newchem-cpp] query inj path names

[mabruzzo]

This should be reviewed after PR #478 has been reviewed and merged

---

The size of the PR grew a little more than I initially anticipated.

Overview

This PR exposes information about the injection-pathway data to users through injection pathway information. Specifically users can query:

• "inject_model_names" to get the list of strings naming each injection pathway (in the order that is internally assumed)
• "inject_path_gas_yield_frac.C", "inject_path_gas_yield_frac.O", "inject_path_gas_yield_frac.Mg", ... to get yields for each metal nuclide that is initially in the gas-phase. For concreteness, let's consider the array associated with "inject_path_gas_yield_frac.C". Element i of that array holds the fraction of all injected non-primordial material corresponding to Carbon nuclides in the gas phase molecules.
• "inject_path_grain_yield_frac.MgSiO3_dust", "inject_path_grain_yield_frac.AC_dust", ... to get yields for each relevant grain species. For concreteness, let's consider the array associated with "inject_path_grain_yield_frac.AC_dust". Element i of that array holds the fraction of all injected non-primordial material initially corresponding to amorphous carbon dust.

All other changes are effectively consequences

C++ Testing

I needed to update a bunch of the C++ tests to make sure that they worked properly since "inject_model_names" is the very first queryable-key that corresponds to an array of strings.

I also added a handful of tests to explicitly test queries of these new keys. Some examples include:

• testing that all of these arrays have the same length
• checking that all yields have a value $ 0 \leq x \leq 1$
• that the sum of all yields is less than or approximately equal to 1 (this is performed separately for different injection pathways)

I did a little refactoring to try to make this all work nicely

gracklepy changes

I updated FluidContainer (and the supporting functions) to explicitly query the list of injection pathways from Grackle, rather than hardcoding the names of all of the injection pathways

• this isn't a significant change, but I very much consider this to be a sign of things to come (i.e. reducing duplication of logic between the core library and the gracklepy layer)

I also added some experimental methods to access the yields from gracklepy. I don't think these are necessarily optimal, but I think its a useful short-term solution. Eventually (but not in this PR), we probably want to reuse this information to improve tests of multi-grain-species dust model so that it appropriately uses this information

pytest changes

I added a simple check of the new gracklepy changes (I think it will be a useful diagnostic tool if we make changes and forget to fully propagate the changes throughout the gracklepy)

The payoff

A major payoff to all of this effort is actually coming in the next PR.