Passes 3003–3011: M36 symmetry, 28-check D4 routing, and typed golden control

[wilcompute]

Scope

Executes all five requested continuations and two additional outside-box fronts.

Exact completed results

• 3003 — M36 symmetry quotient: the six non-CSS success-1/27 stabilizer-output hits from the 649,940-subspace pilot are one orbit of size 729 under Stab(m)^3 ⋊ S3, order 34,992. The 495 RREF pivot patterns reduce to 98 copy-permutation types. The separate 495-shard exhaustive census has been launched but is not claimed complete.
• 3004 — improved D4 localization: a new 28-triangle schedule separates all 48,826 no/one/two-edge nonidentity D4 fault hypotheses, strictly improving the prior 29-triangle construction. Minimum optimality remains open in [23,28].
• 3005 — typed golden hardware: synthesizable A4 syndrome-shell RTL uses the Fibonacci mod-2 action and an 89/233 Christoffel scheduler. Shell operations have no write path into the protected D4 core; core operations require explicit authorization.
• 3006 — objectwise 540×12 bridge: each of the 540 skew isotropic line pairs lies in exactly three spreads and has four isotropic transversals, giving a PSp-transitive 6,480-flag bundle. Both order-12 PGSp classes contain six-cycles, so no projective automorphism realizes the controller's semiregular 12^540 clock.
• 3007 — component sequential receiver: component counts feed a three-outcome Bayesian chirality channel and optimal stopping within the prior-conditioned Helstrom family. Under the synthetic repository stack, copy cost 1e-3 yields modeled success 0.996589 at 7.710 expected copies.
• 3008 — BONKERS tetrahedral shell: the four V4 labels carry the tetrahedral A4 information geometry and an exact visibility-dependent SIC readout model.
• 3009 — BONKERS reversible calendar: the 89/233 Christoffel scheduler crossed with D12 gives a 2,796-tick calendar, 89 events in every phase, global gaps {2,3}, and reversal by cyclic shift 89.

Source, RTL, manuscripts, and evidence

• Adds exact generators/certificates, focused pytest, A4/D4 RTL and testbench, a four-front-door integrator, and a dedicated RTL/PDF evidence lane.
• Adds a separate 99-batch workflow covering all 495 duplicate-free M36 shards and a fail-closed aggregator that accepts completion only at exactly 213,648,435 examined subspaces.
• Adds shared inserts for w33_paper.tex, photonic_holonet.tex, holonet_machine_blueprint.tex, and docs/index.html.

Merge gate

Do not merge until the focused workflow has regenerated all certificates, passed the RTL simulation/synthesis checks, integrated the four front doors, and compiled all three manuscripts. The full M36 census is a separately typed exhaustive job and is not falsely treated as complete while queued.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[wilcompute]

Local exact execution completed before publication:

• focused Python regression: 8/8 passed;
• all six general-isotropic M36 pilot hits collapse to one 729-element orbit under the 34,992-element three-copy deep-ray symmetry;
• the 28-triangle schedule is injective on all 48,826 no/one/two-edge nonidentity D4 hypotheses;
• the 6,480 skew-pair/spread/transversal flags are one PSp(4,3) orbit, while neither PGSp order-12 class has cycle profile 12^540;
• the A4 model, component-channel stopping policy, tetrahedral shell, and 2,796-tick calendar recompute exactly.

The focused GitHub lane is currently queued (run 30865461680). It will merge the newest master, then run Icarus/Yosys, regenerate all certificates, integrate the four front doors, and compile the three manuscripts. The 495-shard M36 census is separately fail-closed and is not being claimed complete while its runner state is unobserved.