Procedural code generation

[jmason42]

Several files (equilibrium_odes.py, two_component_system_odes.py, two_component_system_odes_fitter.py, metabolism_constraints.py, others) written by one piece of upstream code (write_metabolic_constraints_file.py, write_ode_file.py) and executed by another (where?). This is confusing to both humans and code inspection tools, largely undocumented, and its relationship with version control is unclear.

[1fish2]

See #111 for one approach. In brief: lazily load the saved data, generate the code for a lambda, and eval() that (or generate a def and exec() that).

Can Numba JIT-compile generated code without writing it to a file?

@jmason42 offered another approach: lazily load the saved data and interpret it with a Cython function that embodies the rest of the code-generation.