Update the `GGUF usage with llama.cpp` doc page

[Julien Chaumond]

there's now way cleaner syntax than https://huggingface.co/docs/hub/en/gguf-llamacpp to load a model from HF in llama.cpp