Update DORT to follow PyTorch changes

[wschin]

Fix #16355. The root cause change in PyTorch is #103302, which seem blocking calling make_fx inside a dynamo backend.

Changes:

1. Move decomposition to register_backend.py, so we don't have to call make_fx inside DORT, which triggers a bunch of new exceptions.
2. Remove shape inference based on FakeTensorProp since the FX graph received from dynamo contains all shapes now.
3. Fix a macro bug so that DORT can build without CUDA.

Before (3),

#if defined(USE_CUDA) || defined(USE_ROCM)
  virtual PhiloxGenerator& PhiloxGenerator__Default() = 0;
#ifdef ENABLE_TRAINING_TORCH_INTEROP
...
#endif
#endif

After (3),

#if defined(USE_CUDA) || defined(USE_ROCM)
  virtual PhiloxGenerator& PhiloxGenerator__Default() = 0;
#endif
#ifdef ENABLE_TRAINING_TORCH_INTEROP
...
#endif

The later one looks better since the ENABLE_TRAINING_TORCH_INTEROP is for Python bridge code, not for random-number-generating kernels PhiloxGenerator.

[askhade]

Please update the PR description to explain the fix. It will help to review the code as I am not familiar with the code under torchdynamo dir.

[wschin]

Please update the PR description to explain the fix. It will help to review the code as I am not familiar with the code under torchdynamo dir.

Sure. It's done now.