refactor(ata): refresh fixed-form Rust ownership RED on current main

[seonghobae]

Clean-lineage replacement for #750 / issue #629

Protected main advanced after #750 was created. Fresh compare proves #750 is three protected-main commits behind and its only unique delta is tests/test_test_form_rust_ownership.py. This Draft recreates that exact intentional public-boundary RED directly on current protected main; predecessor checks/reviews do not transfer.

Exact identity

• protected base: 8a56d7b1b024e18a5a96a94db4ede1bf6648ead1;
• exact source head: e8cd669de25ea5344b82890694ddc6d2b3bd55a2;
• compare to protected main: direct one-commit descendant, ahead 1 / behind 0;
• changed path: tests/test_test_form_rust_ownership.py only.

Intentional RED

The test replaces proposed _core.assemble_test_form_greedy with a sentinel valid form and requires public assemble_test_form() to return the Rust-owned selection while leaving caller information/content/exclusion arrays unchanged. Current Python-owned ordering/content-feasibility logic should fail the delegation assertion only after normal Rust/PyO3 package setup. Setup/import/fixture/changelog failure is not valid RED evidence.

Minimum GREEN

Move descending finite-information ordering, exclusion filtering, deterministic tie policy, content minima/maxima and look-ahead feasibility into one bounded mlsirm-core kernel. Expose thin PyO3 marshalling; Python retains bounded type/shape/control validation, canonical content-label marshalling and result transport only. Missing/incompatible compiled capability must fail closed. Preserve current deterministic semantics and stable infeasibility behavior; add direct parity/resource/adversarial/immutability evidence, complete docs/rustdoc and changelog only after GREEN, then one unchanged final head with Python 3.12/3.14, Rust/PyO3, package/reinstall, existing GPU/fuzz, Security/SAST and current-head governance gates.

No stronger ATA optimizer, provider, hosted state, dependency/workflow/version/release or competing canonical docs belongs here. Advances #629. Once this replacement is established, #750 is superseded.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 3916a2b7-5476-42f2-b6f7-411ed8ecf0c5

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 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.}

[seonghobae]

Superseded by current-main production implementation PR #753.

Fresh comparison shows #753 is a direct current-main implementation slice that already contains the same public ownership sentinel and caller-immutability assertions from this test-only RED, plus the Rust core kernel, PyO3 binding, public Python delegation, Rust unit coverage, doctoring and changelog material. The only textual delta in this PR's test is the module docstring prefix (Fail-first versus the implementation branch's Ownership contracts); it carries no distinct verification behavior or product value.

Do not transfer checks/reviews between heads. #753 exact head fadf35aa00e09a533010e228c9bdca9ee653e2eb is the sole implementation/acceptance path for this fixed-form assembly concern. Closing this redundant test-only branch avoids two owners for one concern.