[Feature] CREATE EXTENDED guide on how and what torch and python version to choose? (and cuda toolkit)

### Feature Idea

The problems comes with the fact that doing: pip install torch.. t.. does not always get you the same version, I mean sometimes you can force the dev latest and nightly version, sometimes it get you a lower version depending on your py version
sometimes you should not need to update your comfy to avoid py upgrade, but you need a higher torch version anyway, see?

See how all confusing it is?

The more confusing it was for me when I read this chain of comments: https://www.reddit.com/r/StableDiffusion/comments/1h7hunp/comment/m0n6fgu/
How is it possible that someone solved an error by reverting to py 3.11, while the other solved the SAME error by upgrading 3.12?

Can you explain?
And.. I want a guide that everybody can read learn (can be updated by people with contributions), but you need to start it
it will include every possible combinations and why I should choose this py this torch et to solve this or that problem, and why I should or should avoid updating comfy etc etc.
This is a must at this point.

### Existing Solutions

_No response_

### Other

_No response_