Update export/export-aot-inductor benchmark code (#106323)
Update export/export-aot-inductor benchmark code to use recent changes related to kwarg inputs and dataclass outputs.
Updated [dashboard](https://hud.pytorch.org/benchmark/compilers?startTime=Mon%2C%2031%20Jul%202023%2017%3A28%3A05%20GMT&stopTime=Tue%2C%2001%20Aug%202023%2017%3A28%3A05%20GMT&granularity=hour&suite=torchbench&mode=inference&dtype=bfloat16&lBranch=angelayi/benchmark&lCommit=f0987867a88b0b9510fcaf33307150e61517e7a1&rBranch=main&rCommit=f23d755e1f835485b8fef5661e7f983b520d844e)
80% pass rate on HF for export: P801372961
20% pass rate on HF for export-aot-inductor: [link](https://hud.pytorch.org/benchmark/huggingface/inductor_aot_inductor?startTime=Mon,%2031%20Jul%202023%2017:08:02%20GMT&stopTime=Tue,%2001%20Aug%202023%2017:08:02%20GMT&granularity=hour&mode=inference&dtype=bfloat16&lBranch=angelayi/benchmark&lCommit=f0987867a88b0b9510fcaf33307150e61517e7a1&rBranch=main&rCommit=f23d755e1f835485b8fef5661e7f983b520d844e)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/106323
Approved by: https://github.com/desertfire