llama.cpp

common/chat : tighten bare function parsing for Qwen models

#26793

Merged