[Attention][Perf][Kernel] Replace torch.cat with vectorized CUDA kernel MLA query concat - DeepSeek-V3.2#34917

vllm-bot merged 17 commits into

vllm-project:mainvllm-project/vllm:mainfrom

LopezCastroRoberto:concat_mla_qLopezCastroRoberto/vllm:concat_mla_qCopy head branch name to clipboard