feat: add normalization techniques

[EyaRhouma]

Add 3 new techniques for image normalization:

• Local contrast normalization module: inspired by computational neuroscience, aims to make the features in feature maps more significant. Paper
• Local response normalization module: Also inspired by computational neuroscience, aims to reduce the activity of its neighbors and refers to the capacity of a neuron Papers
• Image smoothing module: helps reduce the noise using OpenCV smoothing techniques (gaussian, averaging, median, bilateral)

[JoanFM]

Hey @EyaRhouma ,

this looks like a very cool feature.

But I think it may be much better to implement this logic in a different hub executor that can apply torch transformations inspired in https://github.com/jina-ai/jina-hub/tree/master/crafters/image/ImageTorchTransformation

This executor has not been yet ported to 2.0 and our new hub. But maybe we can try to do so

[EyaRhouma]

Thank you for your fast response. I completely agree ImageTorchTransformation seems to be a better fit to add the new normalization techniques.
I will move everything there.