burn_cell_[metal,primordial]_chem redefines primary_species_* to be real

In the top-level unit_test/_parameters, we have:

primary_species_1  string   ""
primary_species_2  string   ""
primary_species_3  string   ""

the intent of this is to set the name of the dominant species, and then create various mixtures of the composition including these species
using the setup_composition() function. This is used, e.g., by test_react to create a number of different compositions that are mapped onto one of the dimensions of the thermodynamic data cube we are testing with.

burn_cell_primordial_chem redefines these as real:

primary_species_1            real       1.0e0
primary_species_2            real       0.0e0
primary_species_3            real       0.0e0
primary_species_4            real       0.0e0
primary_species_5            real       0.0e0
...

this is a bad design.

I think that burn_cell_primordial_chem and burn_cell_metal_chem should instead use the parameters:

X1            real       1.0e0
X2            real       0.0e0
X3            real       0.0e0
X4            real       0.0e0
X5            real       0.0e0
X6            real       0.0e0
X7            real       0.0e0
...

that are already defined in unit_test/_parameters or use a different name altogether so it doesn't clash with the convention used by other problems.