llmux 0.3.1

Zero-reload model switching for vLLM - manages multiple models on shared GPU
Documentation