Add torch.compile escape hatches to programming model doc

[ydwu4]

As titled.

cc @williamwen42 @msaroufim @anijain2305 @angelayi

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3792

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure

As of commit f00a68f with merge base 29580ff ():

NEW FAILURE - The following job has failed:

• Build tutorials / build / pytorch_tutorial_build_worker (2, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  RuntimeError: f2() Expected a value of type 'Tensor (inferred)' for argument 'y' but instead found type 'int'.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[williamwen42]

Sorry for the late review! I think this should go into the primary programming model docs, not the tutorial.

e.g. mention leaf_function as an escape hatch here: https://docs.pytorch.org/docs/stable/user_guide/torch_compiler/compile/programming_model.fullgraph_true.html#strategy-2-pure-functions-can-always-be-compiled-via-an-escape-hatch

And then write a new page for leaf_function such as https://docs.pytorch.org/docs/stable/user_guide/torch_compiler/compile/programming_model.dynamo_nonstrict_trace.html