ns for fx: make torch.ops.quantized.dropout mapping dynamic
Summary:
Instead of hardcoding the relationship between `F.dropout` and `toq.dropout`,
read it from the mapping.
The mapping itself might need to be in the lowering file, but that's a separate
issue.
Test plan:
```
python test/test_quantization.py -k FXGraphMatcher
python test/test_quantization.py -k FXNumericSuite
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/76989
Approved by: https://github.com/jerryzh168
