CI Benchmark

Update [04/08/2025]: We are currently working on reorganizing the compute for performance benchmark. In the meantime, please reference PyTorch Hub's Dashboard for vLLM: https://hud.pytorch.org/benchmark/llms?repoName=vllm-project/vllm