An exact penalty function approach to all-time-step constrained discrete-time optimal control problems