Transforms v2 documentation not congruent w/ fn parameters (```padding_mode```)

📚 The doc issue

In many geometric transformations (e.g.: ElasticTransform, RandomAffine, ...) there's a fill parameter specified, and its description states:

Pixel fill value used when the padding_mode is constant [...]

Issue is, the padding_mode parameter does not exist in the majority of these transforms' constructors! It's a very minor nitpick, but still 😉

Suggest a potential alternative/fix

The quickest fix is obviously removing the part that says 'when the padding_mode is constant', but it could also be turned into a very fun feature request!

A way to implement this functionality would be to just add a quick computation to find out the image size necessary to cover all the frame, perform the Pad transform, and finally complete the geometric transform. There might be a few edge cases though, so that could possibly require some testing.

Start with the linked ElasticTransform and RandomAffine documentation pages, then compare each fill description with the constructor parameters. Decide whether the documentation should remove the padding_mode condition or whether a feature is intended; done means the documented parameters and behavior are consistent. No file or test path is identified in the issue.