Cherrypick: nccl ops multi gpu #2520

Job	Run time
generate-matrix / generate	10s
filter-matrix	5s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script.sh, packaging/env_... / manywheel-py3_9-cuda11_8	5m 39s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script.sh, packaging/env_... / manywheel-py3_9-cuda12_4	6m 4s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script.sh, packaging/env_... / manywheel-py3_9-cuda12_6	6m 4s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script.sh, packaging/env_... / upload / manywheel-py3_9-cuda11_8	11s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script.sh, packaging/env_... / upload / manywheel-py3_9-cuda12_4	15s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script.sh, packaging/env_... / upload / manywheel-py3_9-cuda12_6	13s
Test dynamo converters [Python]	0s
Test dynamo frontend [Python]	0s
Test dynamo core [Python]	0s
Test core [Python]	0s
Test dynamo cudagraphs [Python]	0s
Test dynamo export serde [Python]	0s
Test torch compile backend [Python]	0s
Test torchscript frontend [Python]	0s
	18m 41s

Provide feedback