🐛 Make QTensor shrinking sparse and atomic

[simon1hofmann]

🤖 AI text below 🤖

Description

Makes qtensor-shrink-to-fit safe for very large declared tensors and
unsupported non-linear tensor chains.

The rewrite previously allocated dense helper vectors proportional to the
declared tensor size and asserted that every tensor value had exactly one use.
It now builds a sparse, read-only access plan, validates the complete chain
before rewriting, and commits the compact replacement only after planning
succeeds.

This is one focused replacement for draft PR #2287 and addresses part of #2255.

Validation

• full QTensor transform suite — 3/3 passed
• uvx nox -s lint — passed
• git diff --check — passed

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.
• I have added entries to the changelog for any noteworthy additions, changes, fixes, or removals.
• I have added migration instructions to the upgrade guide (if 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.

AI assistance disclosure: Codex extracted and simplified this focused change
from the authorized #2255 audit, validated it, and drafted this description.

[burgholzer]

This LGTM except for one small detail

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[coderabbitai]

Important

Approval pending

CodeRabbit has no unresolved comments, but it has not reviewed the latest commit.

Use the checkbox below to review the latest commit. CodeRabbit will approve the changes if it finds no blocking issues.

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}