[Perf] Set Flashinfer sparse MLA as default backend for FP8 kv cache#37252

MatthewBonanni merged 3 commits intovllm-project:mainvllm-project/vllm:mainfrom

wzhao18:wzhao/update-sparse-mla-prioritywzhao18/vllm:wzhao/update-sparse-mla-priorityCopy head branch name to clipboard