Operant matching is not a logical consequence of maximizing reinforcement rate