Fix main red: clippy, stale QEC snapshots, GHZ lattice-surgery reset bug

[arniber21]

What's in this PR

cargo test --workspace (and clippy) were red on main. Three independent issues:

1. Clippy: 7 unused-import errors in quon_na/src/movement/{duals,emit,geometry,plan}.rs, left over from the NA: Carve movement / zoned / compaction monoliths by interface #215 movement-monolith carve-up.

2. Stale QEC report snapshots: the accepted .snap goldens for qec_repetition_d3_hybrid/qec_surface_d3_hybrid predated QEC: implement magic-state-consuming logical T and CCZ operations #283/QEC: magic-state-consuming logical T and CCZ operations (#283) #288's new resource-report fields (t_count, tdag_count, ccz_count, magic_state_demand) and never got re-accepted, so cargo test -p quon_na failed. Also found two .snap.new insta scratch files tracked in git by mistake (from that same PR) — insta pending-snapshot files are meant for local review via cargo insta review, never committed. Untracked them and added *.snap.new to .gitignore.

3. Real correctness bug: quonc/tests/samples_catalog.rs::ci_smoke_entries_typecheck_with_quonc failed on the new GHZ benchmark sample (samples/neutral-atom/benchmarks/surface_d3_ghz.qn, from Samples: reconfigurable neutral-atom QEC benchmark suite #284/Samples: reconfigurable neutral-atom QEC benchmark suite (#284) #292) with quantum.na verification failed: cycle 425: atom 36 is reused ... after measure at cycle 346 without an intervening reset.

   Root cause: quon_qec/src/patch_ops.rs's MeasureAncilla handling (the live logical-CX lattice-surgery lowering) measures only the ancilla patch's top-row atoms when a logical CX finishes, but never resets any of the ancilla's atoms — unlike the seam-atom measure+reset pairing rough_merge_round/smooth_merge_round/split_seam_round already do correctly. Chaining two logical CX ops (as the GHZ sample does: logical_cx(a,b) then logical_cx(b,c)) leaves the first ancilla's atoms parked, un-reset, in the trap grid; the verifier correctly catches it the first time the second CX's movement/geometry planning has to move one of those atoms out of the way.

   Fixed by resetting the ancilla's full atom footprint (all data + check atoms, not just the measured top row — by the Horsman gadget's design the ancilla is fully disentangled from control/target by this point, so this is physically sound) right after the logical-Z measurement. Applied the identical fix to the lattice_surgery.rs reference-only path for consistency. Updated the two tests whose terminal-length assertions assumed measure-only (quon_qec/src/expand.rs, quon_qec/tests/patch_ops.rs).

Verification

All run locally, not just asserted:

• cargo fmt --check — clean
• cargo clippy --workspace --all-targets -- -D warnings — clean
• cargo test --workspace — 0 failures (confirmed 3x, including a full rebuild)
• .venv/bin/python -m pytest python -q — 78 passed, 1 skipped
• pnpm --dir website build — clean
• Reproduced the original GHZ failure with the pre-fix binary, then confirmed the post-fix binary compiles the same sample cleanly (quantum.na verification passed).

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
quon	Ready	Preview, Comment	Jul 20, 2026 8:25pm

[arniber21]

• Fix main red: clippy, stale QEC snapshots, GHZ lattice-surgery reset bug #313 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.