Experiences with a space-time multigrid method for the optimal control of a chemical turbulence model