26 Polishing DSRGAN notebook

[mxochicale]

This PR address #26.

List of points that will help to improve the reproduciblity of the notebook

• [x] Diffusion Model This will take X days to train.
• [x] DSGAN This will take a X hours to complete with X GPU.
• [x] Reference of degradation
• [x] (x/2) What's the meaning of 127.5 in
  • aug_images = (aug_images[1] + 1) * 127.5
  • original_images = (original_images[1] + 1) * 127.5
  • colour_key = (colour_key + 1) * 127.5 and
  • sr_img = (sr_img + 1) * 127.5
• [x] 64 value appears in the notebook and something is confusion if the value is from train_size or half of the size of the original image.
• [x] Why 192? in fid = FrechetInceptionDistance(feature=192).to('cuda') > https://github.com/budai4medtech/xfetus/pull/33#discussion_r1273304484
• [x] How to use www.wandb.ai for FID score (optional)
• [x] we might like to move all the variables at the top of the script and perhaps put a yaml file but let's discuss this one. (Optional)

[mxochicale]

Hi @harveymannering I have re-based this branch and added few minor changes to the notebook, let me know if you have any suggestions otherwise, we might like to merge it next week.

[harveymannering]

These changes all look good to me. Thanks for re-basing!