llmux 0.2.1

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