Extension of the TVD Midpoint Scheme to Higher-Order Accuracy in Time