Differentiable JPEG function wrong positional argument name

[Fleyderer]

Describe the bug

kornia.enhance.jpeg_codec_differentiable() requires input positional argument while there is no such argument in function

I know that it is because of @perform_keep_shape_image decorator, but this behaviour is not normal

And yes, it is working if I change image_rgb=a to input=a

Reproduction steps

1.

 pip install git+https://github.com/kornia/kornia

2.

import torch 
import kornia

a = torch.rand(4, 3, 112, 112)
kornia.enhance.jpeg_codec_differentiable(
    image_rgb=a, 
    jpeg_quality=torch.full([a.shape[0]], 15))

3.

TypeError: jpeg_codec_differentiable() missing 1 required positional argument: 'input'

Expected behavior

Expected behaviour is to have ability of using named arguments in function

Environment

I think it does not matter in this case

Additional context

No response

[johnnv1]

this seems an issue related to the decorator perform_keep_shape_image cc @edgarriba @ChristophReich1996

kornia/kornia/utils/image.py

Line 214 in d6e1485

def perform_keep_shape_image(f: Callable[..., Tensor]) -> Callable[..., Tensor]:

[edgarriba]

Ideally we rename to input and more ideally as a meta task we standardise that across the library /cc @johnnv1