llama.cpp

presets : add qwen3-30B-a3b FIM

#15616

Merged