[#7588][feat] lock gpu clocks in test_perf.py to reliably detect perf regressions#8099

MrGeva merged 2 commits intoNVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

nv-auto-deploy:lock_gpunv-auto-deploy/TensorRT-LLM:lock_gpuCopy head branch name to clipboard