[quant][graphmode][fx] Add support for specifying reference quantized module mapping in backend_config_dict (#68227)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/68227
This PR adds two keys to backend_config_dict:
"root_module": the root module for the pattern (since we may have patterns for fused ops)
"reference_quantized_module_for_root": the corresponding reference quantized module for the root
Test Plan:
```
python test/test_quant_trt.py TestQuantizeFxTRTOps
python test/test_quant_trt.py TestConvertFxDoNotUse
```
Imported from OSS
Reviewed By: vkuzo
Differential Revision: D32537711
fbshipit-source-id: 6b8f36a219db7bb6633dac53072b748ede8dfa78
