Lossless convexification of control constraints for a class of nonlinear optimal control problems