dbr quant overhead[12/x]: turn off overrides for module convert output hook (#68839)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/68839
We can assume that there are no overrides needed for the hook which
dequantizes the module outputs, so we can turn them off explicitly.
While this does not lead to a measurable perf win, it makes things
easier to debug by eliminating the no-op overrides.
Test Plan:
```
python test/test_quantization.py TestQuantizeDBR
```
Reviewed By: jerryzh168
Differential Revision: D32630886
Pulled By: vkuzo
fbshipit-source-id: 1719c168f5f21f3e59c80a3b6d0f32ebb1c77ef8