[Torch.Export] Test Dynamic Shape Originated from 0/1 Static Shape

If an axis in the example input has size 1, then torch.export refuse to make that axis dynamic due to 1 being special in broadcast. PyTorch team is actively investigating, see [their discussion](https://docs.google.com/document/d/16VPOa3d-Liikf48teAOmxLc92rgvJdfosIy-yoT38Io/edit?fbclid=IwAR3HNwmmexcitV0pbZm_x1a4ykdXZ9th_eJWK-3hBtVgKnrkmemz6Pm5jRQ#heading=h.ez923tomjvyk)

Once they have the 0/1 shape static -> dynamic issue figured out, we will test our converter accordingly

Start with the linked PyTorch discussion about torch.export dynamic shapes and inspect the repository's converter tests. Once torch.export supports converting axes with static size 0 or 1 to dynamic, add coverage for the converter; done means the relevant conversion tests pass for those cases.

Written by the indexing model from the issue text.