[Datasets] [AIR] Add TorchVisionPreprocessor

[bveeramani]

Description

Add a preprocessor like

TorchVisionTransformer(transform: Callable[[], torch.Tensor], columns: list[str])

that applies TorchVision transforms.

Use case

To apply a simple TorchVision transform like:

transform = Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) 

You need to something like this:

def map_fn(batch: dict[str, np.ndarray]) -> dict[str, np.ndarray]:
    return {"image": np.array([transform(image).numpy() for image in batch], dtype=object)}

which is complicated.

With a TorchVisionTransformer, you could instead do:

TorchVisionTransformer(transform, columns="image")

which is simpler.