Add components to perform super-resolution based on diffusion models

[Warvito]

Create the components necessary to perform the training and inference of super-resolution task based on diffusion models. As reference, we could use approaches used in Imagen (https://imagen.research.google/), Cascated Diffusion (https://cascaded-diffusion.github.io/), or ediff-i (https://deepimagination.cc/eDiff-I/).

[Warvito]

SR3+ (https://arxiv.org/abs/2302.07864v1)