[reland][quant][graphmode][fx] Merge all quantization mode (#45292) (#45672)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/45672
This PR merges all quantization mode and will only expose the following top level functions:
```
prepare_fx
prepare_qat_fx
convert_fx
```
Test Plan:
Imported from OSS
Imported from OSS
Reviewed By: z-a-f
Differential Revision: D24053439
fbshipit-source-id: 03d545e26a36bc22a73349061b751eeb35171e64
