vllm

[Linear Attention] fix bug for linear attention + prefix caching + reset_prefix_cache

#35157

Merged