### What this PR does / why we need it? Add layernormFn triton op for qwen3Next model for better performance. <img width="248" height="526" alt="image" src="https://github.com/user-attachments/assets/27b47157-5df5-4db1-aa88-1dae799b2bf6" /> ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? - vLLM version: v0.13.0 - vLLM main: https://github.com/vllm-project/vllm/commit/2f4e6548efec402b913ffddc8726230d9311948d --------- Signed-off-by: SunnyLee219 <3294305115@qq.com>