change operator_range to work with lower and upper in op bench (#30096)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/30096
as title
Test Plan:
```
buck run mode/opt caffe2/benchmarks/operator_benchmark:benchmark_all_quantized_test -- --iterations 1 --operator_range a-a
# ----------------------------------------
# PyTorch/Caffe2 Operator Micro-benchmarks
# ----------------------------------------
# Tag : short
# Benchmarking PyTorch: add
# Mode: Eager
# Name: add_N2_dtypetorch.quint8_contigTrue
# Input: N: 2, dtype: torch.quint8, contig: True
Forward Execution Time (us) : 22.251
# Benchmarking PyTorch: add
# Mode: Eager
# Name: add_N2_dtypetorch.qint8_contigTrue
# Input: N: 2, dtype: torch.qint8, contig: True
Forward Execution Time (us) : 17.247
# Benchmarking PyTorch: add
# Mode: Eager
# Name: add_N2_dtypetorch.qint32_contigTrue
# Input: N: 2, dtype: torch.qint32, contig: True
Forward Execution Time (us) : 29.653
...
Reviewed By: hl475
Differential Revision: D18596447
fbshipit-source-id: eac8d9d90db244aa9799293c22bb0d30cf3edf58
pytorch
23991e89 - change operator_range to work with lower and upper in op bench (#30096)
