[FEATURE REQUEST] Reduction of memory for HISTORY diagnostics with species dimensions

[yantosca]

Currently, the State_Diag arrays for HISTORY diagnostics with species dimensions (e.g. SpeciesConc, DryDep, WetLossConv, WetLossLS) are always allocated with the maximum possible number of slots (e.g. State_Chm%nSpecies, State_Chm%nDrydep, etc.). This can cause a large amount of memory to be allocated, especially if only a subset of species need to be archived.

@yantosca is currently working on an update that will only allocate as much memory as there are species requested in HISTORY.rc. For example, if you only request SpeciesConc_O3 and SpeciesConc_NO, then the State_Diag%SpeciesConc array will only be allocated with (NX, NY, NZ, 2) slots instead of (NX, NY, NZ, State_Chm%nSpecies) slots. This will result in a substantial savings of memory and also CPU cycles.

This work, which leverages the DiagList and TaggedDiagList data structures implemented by @christophkeller and @lizziel, should be ready for 12.9.0. More information will be forthcoming soon.

[yantosca]

Also need to handle the case if a diagnostic has more than one wildcard.

[yantosca]

This has now been completed. Initial testing of the budget diagnostics shows a ~60% decrease in the time needed to compute budget diagnostics for only 2 species vs. budget diagnostics for all species. Further time testing is needed.

This update spans several commits, but the most recent are f847746 and geoschem/GCClassic@ed800c8.