🐛 Reject nonlinear QTensor shrink chains safely

[simon1hofmann]

🤖 AI text below 🤖

Summary

• Require each tensor in a shrink chain to have exactly one user before dereferencing it.
• Leave nonlinear or incomplete chains unchanged.
• Add a no-mutation regression for a tensor with two deallocation users.

Follow-up to #2295 and part of #2255.

Validation

• QTensor transforms unit-test suite: 3 passed.
• uvx nox -s lint: passed.
• uvx nox -s cpp-lint: could not start locally because clang-tidy 22 is unavailable; CI will run the canonical check.

AI assistance: Codex extracted the minimal follow-up from the contract audit branch and ran the listed validation.

Checklist

• The pull request only contains commits that are focused and relevant to this change.
• I have added appropriate tests that cover the new/changed functionality.
• I have updated the documentation to reflect these changes. (Not applicable: no user-facing documentation change is needed.)
• I have added entries to the changelog for any noteworthy additions, changes, fixes, or removals. (Not applicable: this focused fix is labeled skip-changelog.)
• I have added migration instructions to the upgrade guide (if needed). (Not needed.)
• The changes follow the project's style guidelines and introduce no new warnings.
• The changes are fully tested and pass the CI checks.
• I have reviewed my own code changes.

If PR contains AI-assisted content:

• Any agent that created, edited, or submitted GitHub content was explicitly authorized for that scope, as required by our AI Usage Guidelines.
• Every agent-authored or agent-edited public text body begins with the visible disclosure 🤖 *AI text below* 🤖 (titles are exempt).
• I have disclosed AI assistance in the PR description.
• I confirm that I have personally reviewed and understood all AI-generated content, and accept full responsibility for it.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[burgholzer]

This feels like a fix at the wrong level. QTensor, by-definition, is a linear type. Any violation of that constraint already constitutes an invalid program. Please revisit.

[simon1hofmann]

🤖 AI text below 🤖

Agreed. This follow-up reintroduced the same one-use guard and invalid nonlinear regression that were removed from #2295 after review. QTensor linearity is the owning contract, and qco::verifyLinearity already requires each tensor value to have exactly one use. I am closing this PR as redundant.