Add Llamaport to User Interfaces

[smkamranqadri]

Adds Llamaport, an open-source macOS GUI for llama-server.

It reads GGUF headers directly, so quantisation, context length and MoE structure come from the file rather than the filename. The full launch command is shown before a model starts, and while it serves the app reports KV cache, generation and prompt-eval rates, memory pressure and swap. Downloads from Hugging Face use ranged segments, resume after a kill, verify sha256 and queue.

It deliberately has no chat interface of its own — a running model opens llama.cpp's own web UI in a second window — so it sits alongside the chat frontends here rather than competing with them.

Rust and Tauri, MIT. Disclosure: I am the author.