Update

Queued local AI requests by priority and model

Built an Ollama gateway that turns requests into inspectable jobs, orders them by priority and model, and reports their lifecycle and resource state.