Re-land #283 unlocks (Quantum, Disambiguator, verb_table, animal-farm harness) — orphaned by merge order

[AdaWorldAPI]

Why this PR exists

PR #283 was marked merged but the unlock content never landed on main. The merge order on 2026-04-29 was:

time	event
06:26:38	#282 (grammar-fixes) merged to main
06:26:45	#283 (grammar-unlocks) merged to claude/grammar-fixes-r2-2026-04-29 (its base — already drained into main)

#283 landed on the orphaned branch tip 7d7b7224, which is no longer reachable from main. The three unlock commits sit there with nowhere to go.

This PR brings that content forward against the current main. No content change vs the original #283 — same three commits, same +845 / -22 footprint across 9 files.

What's wired

• B-unlock-yaml-style (bbb9898) — YAML robustness + Instrument variant + persistence + 144-cell verb taxonomy table
• B-unlock-bridge (30e43f8) — Trajectory audit-hash bridge + generalised Disambiguator trait + Quantum mode (PhaseTag + HolographicMode) + Animal Farm forward-validation harness
• B-META R2 unlocks (5f17459) — Wires verb_table + disambiguator + trajectory_audit + quantum_mode modules + the PhaseTag u128::MAX as f32 → infinity overflow fix

Files changed (vs current main)

crates/deepnsm/src/lib.rs                          |   ?   modified
crates/deepnsm/src/quantum_mode.rs                 |   ?   added
crates/deepnsm/src/trajectory_audit.rs             |   ?   added
crates/deepnsm/tests/animal_farm_harness.rs        |   ?   added
crates/lance-graph-contract/src/grammar/disambiguator.rs   |   added
crates/lance-graph-contract/src/grammar/mod.rs     |   8 +  modified
crates/lance-graph-contract/src/grammar/tekamolo.rs|   9 +/-1   modified
crates/lance-graph-contract/src/grammar/thinking_styles.rs | 244 +/-   modified
crates/lance-graph-contract/src/grammar/verb_table.rs      | 120 +    added
9 files changed, 845 insertions(+), 22 deletions(-)

Verification

• git merge --no-commit --no-ff of branch onto current main → "Automatic merge went well", exit 0
• No file overlap with anything that landed between original merge-base (eeddfe9) and current main (218c2cf) — that window is empty (only the merge commit 218c2cf itself, no source-file diffs)
• Original feat: Grammar unlocks — verb table, Disambiguator trait, Quantum mode, Animal Farm harness #283 reported all feature combos pass cargo check and 382 tests green

Cross-link

• Original (orphaned, marked merged): feat: Grammar unlocks — verb table, Disambiguator trait, Quantum mode, Animal Farm harness #283
• Companion foundry-unlocks rebase: Re-land #281 unlocks (PolicyRewriter, DomainProfile) — orphaned by merge order #284

Generated by Claude Code

---

Generated by Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5f1745969f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Encode π from the low 64-bit phase domain

to_angle decodes only the low 64 bits of PhaseTag, but pi() sets the full 128-bit value to u128::MAX / 2; that makes the low half u64::MAX, so PhaseTag::pi().to_angle() resolves to ~TAU (2π), not π. Any caller using pi() as a canonical half-turn phase will get the wrong phase offset and distance behavior. pi() should be derived from the same low-half mapping as from_angle (or set low bits to ~u64::MAX / 2).

Useful? React with 👍 / 👎.