Two main things happened:
- I removed {wrap_key, PythonTensor, pythonkey_trace} from being public
APIs
- I moved all compilation related things to the functorch.compile
namespace. This includes nnc_jit which is now in
functorch.compile.nnc_jit
Concerns:
- nnc_jit was in the functorch namespace for a long time. Should we
leave it there? Are there stakeholders to notify?
* Re-land the compile cache
There were a few minor problems, but the major thing was that
handle_torch_function_no_python_arg_parser wasn't exposed via TORCH_API.
* make autograd ops work
