[New]: diffusion models LoRA training playbooks

[bitgamma]

@danielholanda suggested that I open an issue to propose the idea to turn these guides into playbooks:

1. https://bitgamma.github.io/ai-blog/blog/sd-scripts/
2. https://bitgamma.github.io/ai-blog/blog/z-image/

Both are regarding training LoRA for image generation models on Strix Halo (other models would probably work). One is based on sd-scripts and the other on musubi-tuner (both tools from the same author). These command-line only tools are quite popular and support all commonly used models. Unlike Ostris AI-Toolkit (which is probably the most used LoRA trainer) they support ROCm out of the box, which is why I chose them.

If you think this content is worth adding, I will make a proper PR.

[sreeram-11]

Thank you, @bitgamma, for the wonderful ideas! If you could open a PR that adds this content to the playbooks, that would be great. Our team will review it.

Please be sure to follow the contribution guidelines before submitting: https://github.com/amd/playbooks/blob/main/CONTRIBUTING.md

@bconsolvo, could you please confirm?

[bconsolvo]

You will need to change the images so that they are safe for a professional work environment. Currently, they are not. And please follow the contributing guidelines, yes, including CI tests.