[shoumikhin]

Backport of #4881 to release/2.15.

Depends on the #4839 backport (#4910). This change updates the ExecuTorch guide that #4839 adds, so it is built on top of that backport. Until that one merges, this branch also carries the #4839 commit. Please review only the last commit.

In 2.15, the shortest way to write an ExecuTorch file fails:

torch_tensorrt.save(trt_gm, "model.pte", output_format="executorch")
# ValueError: ... retrace is True, however no inputs specs were provided ...

save() retraces by default, while torch_tensorrt.executorch.export() does not, so the two entry points disagree. With this change save() does not retrace for "executorch" unless asked. Other formats keep retracing by default, and an explicit retrace= still wins.

This is a clean git cherry-pick -x of the merge commit on top of the #4839 backport, with no conflicts. The patch is identical to #4881.

Validation:

• the changed files match feat(executorch): do not retrace by default when saving a .pte #4881 exactly
• pre-commit passes on all changed files
• test_api.py on CPU without Torch-TensorRT installed: 146 passed, the same as on release/2.15. The two new tests need Torch-TensorRT, so they run in CI