chore(bb): refresh pinned Chonk IVC inputs (fix nightly debug build)

[AztecBot]

What

Refresh the pinned Chonk IVC inputs so the Nightly Debug Build goes green again.

This PR carries the ci-refresh-chonk label. The bb chonk input update CI job (barretenberg/cpp/scripts/ci_update_chonk_inputs.sh) regenerates the pinned flows, verifies one small flow proves+verifies, and pushes the updated barretenberg/cpp/scripts/chonk-inputs.hash back onto this branch. The only intended committed change is that hash; the trailing-newline touch here just creates a diff to open the PR (the download/check scripts strip whitespace, and the refresh job overwrites the file).

Root cause

Nightly Debug Build run 26705445627 failed. The build/compile succeeded — the failing step was the test bbapi_tests ChonkPinnedIvcInputsTest.AllPinnedFlows (exit code 1).

The pinned Chonk flows are an S3 tarball keyed by chonk-inputs.hash, last refreshed on 2026-05-21 (2ab01bb56fa). Since then ~311 commits landed on next, including breaking protocol-circuit changes that alter the kernel public-input layout (e.g. message signing + fallback keys #23510, embedding BoundedVec max length in validation requests #23622, protocol-contract demotions). That makes the precomputed VKs / kernel IO baked into the pinned msgpack inconsistent with the current circuits.

Symptoms in the failing flow (CI log):

Pub inputs offset mismatch: 669 vs 673
Does the precomputed vk match with the one generated during accumulation? false
Commitment mismatch: Q_M / Q_C / SIGMA_* / ID_* / ...
Assertion failed: (kernel_return_data_match)
Reason: kernel_return_data mismatch: proof contains { 0x18322739..., 0x0 }
        but kernel_calldata commitment is { 0x...01, 0x...02 }

The +4 public-inputs shift (669 → 673) and the wholesale precomputed-VK mismatch are consistent with legitimate circuit evolution, not a proving-system regression — so the correct remediation is to refresh the pin (per barretenberg/cpp/CLAUDE.md and the chonk-inputs skill), which the circuit-changing PRs should have done via the same label.

Verification

The refresh job runs ChonkPinnedIvcInputsTest.AllPinnedFlows on the smallest regenerated flow (native + bb.js) before committing the new hash, and the nightly will exercise the full set against the refreshed tarball.

---

Created by claudebox · group: slackbot