[Documentation] Better documentation that shows how to load stable diffusion locally

[patrickvonplaten]

Describe the bug

We've gotten much feedback about the bad UX of either having to pass an auth_token to from_pretrained(...) or about the cache being to complicated to figure out how to load the model locally => we need to improve the documentation here.

E.g. - see: #441 (comment)

This includes, but is not limited to, adding documentation to this section: https://huggingface.co/docs/diffusers/using-diffusers/loading

Reproduction

No response

Logs

No response

System Info

main

[patrickvonplaten]

Should be closed by #1174