Learning Foresightful Dense Visual Affordance for Deformable Object Manipulation