Local LLMs

I added a section on hosting local LLM with a focus on hardware, tuning, and quantization.

I ended up getting the M4 Max Studio for the LLM Lab, thinking that I can add capacity via clustering as needed.