Add concrete_args to feature extraction tracing. #3231

Job	Run time
unittests-prototype (3.10, linux.12xlarge, cpu) / linux-job	3m 4s
unittests-prototype (3.10, linux.12xlarge, cpu) / linux-job	3m 4s
unittests-prototype (3.11, linux.12xlarge, cpu) / linux-job	2m 56s
unittests-prototype (3.11, linux.12xlarge, cpu) / linux-job	2m 56s
unittests-prototype (3.12, linux.12xlarge, cpu) / linux-job	3m 9s
unittests-prototype (3.12, linux.12xlarge, cpu) / linux-job	3m 9s
unittests-prototype (3.8, linux.g5.4xlarge.nvidia.gpu, cuda, 11.8) / linux-job	42s
unittests-prototype (3.8, linux.g5.4xlarge.nvidia.gpu, cuda, 11.8) / linux-job	49s
unittests-prototype (3.8, linux.12xlarge, cpu) / linux-job	3m 15s
unittests-prototype (3.8, linux.12xlarge, cpu) / linux-job	3m 15s
unittests-prototype (3.9, linux.12xlarge, cpu) / linux-job	2m 55s
unittests-prototype (3.9, linux.12xlarge, cpu) / linux-job	2m 55s
	32m 9s

Provide feedback