llmux 0.7.8

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