Added template framework for applying transforms to distance maps

[andy-s-ding]

nnUNet currently does not have a framework for loading in distance maps that were calculated offline, though I have a version of nnUNet with a modified dataloader that can do this. For full functionality, distance maps should also be cropped and transformed in a similar fashion to the image and segmentation data. The easiest way to do this is to add another key in the data dictionary called 'dist_map'. If the data has this key, then a transform can call _apply_to_dist_map. Modifying the BasicTransform class is the simplest way to add this functionality to other transforms.

[FabianIsensee]

Distance maps should be treated as regression targets. Or am I missing something here?
Best,
Fabian

[andy-s-ding]

Ah, I didn't realize that they could be treated that way. I didn't see anything in the code base that used regression targets, so I made my own data dictionary key for it.