Automate release branch preparation

[JacobSzwejbka]

Summary

Automate the deterministic branch-cut work so release creation no longer depends on a manual checklist. The normal invocation is DRY_RUN=disabled ./scripts/release/cut-release-branch.sh; the release version comes from viable/strict, and the raw cut is preserved before the prepared release branch is committed and pushed.

Review the cut flow first. It resumes an unfinished cut from the preserved commit even if viable/strict advances, requires the Arm API snapshot on main, and keeps dry runs read-only. Dependency preparation selects the next PyTorch RC, derives torchvision while preserving the independently versioned torchaudio pin, verifies their artifacts, selects the matching test-infra branch, and synchronizes the PyTorch source commit plus vendored c10 headers to the release tag.

Packaging uses an explicit wheel variant on Linux, macOS, CUDA, Windows, and QNN builds. CPU release wheels declare a bounded PyTorch range; CUDA wheels declare the exact matching build and reject a stale public train. Minimal and source installs do not acquire release-wheel metadata. Clean-install checks validate the declared metadata against both torch_pin.py and the intended CUDA row. Release-branch lint already uses the test index for RC dependencies.

Stable dependency finalization remains in follow-up #23289. This PR does not change the current PyTorch pin or ExecuTorch version.

Test plan

• python3 scripts/release/test_prepare_release.py (2 tests)
• pytest collection of the preparation suite (2 tests)
• .ci/scripts/tests/test_cu134_dependencies.py (9 tests, 16 subtests)
• mypy, ufmt, and flake8 on the changed Python files
• shell syntax and git diff --check
• live resolution of the PyTorch v2.14.0-rc1 tag and a pinned c10 raw-content comparison

Authored with OpenAI Codex.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/23287

• 📄 Preview Python docs built from this PR

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

❌ 8 Pending, 2 Unclassified Failures

As of commit 5c4b955 with merge base c5f2b7a ():

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

• Test ROCm AOTI / test-voxtral-realtime-rocm-gfx950-rocm7.2 / linux-job (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  RuntimeError: Command docker exec -t c0f022d9b81217ed84a06799ff5d77caae27ff14ea006be11a30513a6f345758 /exec failed with exit code 2
• Test ROCm AOTI / unittest-rocm-gfx950-rocm7.2 / linux-job (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  RuntimeError: Command docker exec -t 6125cce8b7ab78df619416dc037a1c68b360543d9dce7369b4dde42c22d22de7 /exec failed with exit code 2

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

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.