[polyml] Efficient parallel programming in Poly/ML -- continued