CLUBB does not conserve energy

There is a limiter in physics_update that fixes negative vapor produced by clubb. The limiter changes energy, but there is no energy fix after the limiter. This leads to energy leaks from clubb.

I assume that the limiter also contributes to very infrequent MG2 energy errors and maybe ZM errors.

One could try to put a local, active only in a column, energy fixer in physics_update. Overall, I would think limiters should be a part of the physics package, not an interface, but that is an unlikely change. Maybe, placing a mass-conserving limiter in clubb_intr before brute-force energy conservation is computed would fix it. However, at least 2 more packages, ZM and MG2, might need the same change before the limiter can be removed from physics_update.