upgrade to Qwen3-Coder-30B-A3B-Instruct (MoE, 3B active) - ZeroGPU xlarge

Upload folder using huggingface_hub

initial commit