pt2: make aot_eager backend handle basic float8 operations (#107642)
Summary:
Makes aot_eager backend of torch.compile handle basic float8 operations.
This is useful for float8 training UX.
Test Plan:
```
python test/test_quantization.py -k test_pt2_traceable_aot_eager
```
Reviewers:
Subscribers:
Tasks:
Tags:
Fixes #ISSUE_NUMBER
Pull Request resolved: https://github.com/pytorch/pytorch/pull/107642
Approved by: https://github.com/albanD
