Quantise: apt-install cmake before building llama-quantize

Fix PEP 723 blank line

Quantise safetensors -> GGUF (llama.cpp path)