A Instance augmentation

TSP and CVRP experiments in this paper are following the setup where node locations are randomly sampled from the unit square, i.e., x (0, 1) and y (0, 1). All transformations for the 8 instance augmentation used in the experiments preserve the range of x and y, and therefore the new problem instances generated by these transformations are still valid. For the sake of improving the inference result, however, there is no need to stick to "valid" problem instances that comply to the setup rule, as long as the (near-) optimal ordering of node sequence can be generated. Take, for example, rotation by 10 degrees with the center of rotation at (0.5, 0.5). The new problem instance generated by this transformation may (or may not!) contain nodes that are outside the unit square, but this is okay.