fix(knowledge): tier-0 canonical pattern letters (post-#358 correction; Pattern A is SPO-G design phase, not 'OGIT Triangle shipped')

[AdaWorldAPI]

Critical follow-up to merged PR #358

The Tier-0 Pattern Recognition doc landed in PR #358 (merged) with a non-canonical pattern-letter scheme that mislabels several DESIGN-PHASE patterns as SHIPPED. Most critically:

• Pattern A was labeled "OGIT Triangle (Shipped consumer side)"
• Canonical Pattern A is the SPO-G u32 OGIT slot — DESIGN PHASE — tracked by TD-OGIT-G-SLOT-1

This is dangerous because the same doc tells readers "don't build shipped patterns". A future agent reading it would skip TD-OGIT-G-SLOT-1 work (the foundational SPO-G slot for the entire OGIT-G overlay).

What this PR fixes

Rewrites .claude/knowledge/tier-0-pattern-recognition.md (~22 KB → ~26 KB) with:

1. Prepended CRITICAL CORRECTION block at top of doc explaining the prior mislabeling
2. Canonical TL;DR table using W1 master's A-O letter assignments:
   • A SPO-G u32 slot → DESIGN PHASE (TD-OGIT-G-SLOT-1)
   • B Context Bundle → DESIGN PHASE (TD-CONTEXT-BUNDLE-2)
   • C Generic Bridge → DESIGN PHASE (TD-GENERIC-BRIDGE-3)
   • D Meta-Structure Hydration → DESIGN PHASE
   • E Compile-Time Consumer Binding → DESIGN PHASE (TD-MANIFEST-MODULES-4)
   • F ractor Supervisor → DESIGN PHASE (TD-RACTOR-SUPERVISOR-5)
   • G Best-Practice Thinking Inheritance → DESIGN PHASE
   • H Switchable Cognitive Vessel → SHIPPED (p64-bridge::CognitiveShader)
   • I Implicit Cognition → SHIPPED (CycleAccumulator PR Migrate version_watcher to std::sync; add CycleAccumulator; document Single-Binary Topology #337)
   • J INT4-32D Atoms → DESIGN PHASE
   • K Circular Compilation → ASPIRATIONAL
   • L SPO-Chain Narrative → PARTIALLY SHIPPED
   • M Wave-Particle Bimodal → PRIMITIVES SHIPPED
   • N Fingerprint-as-Codebook → SHIPPED
   • O Phenomenological Memory → SHIPPED
3. Per-pattern recognition sections rewritten with canonical letters
4. Substrate inventory appendix preserves W2's file-recognition insights (WorldModel four-section split, Tokenizer Registry, Persona Runtime, Sensor Bridge, Reranker Lens, Contrastive Learner, Branching/Ghosts, Cronbach, Meaning Axes, Engine variants) as supporting infrastructure for canonical patterns — none of these are themselves canonical patterns A-O
5. Pre-work checklist explicitly references the TD-X rows in TECH_DEBT.md for design-phase work
6. Maintenance section declares pattern letter assignments immutable from W1 master

Why a separate PR

PR #358 merged before this fix could land. Rather than direct-commit to main, this PR exists for visibility — the tier-0 doc is the load-bearing entry point for all future architectural work; the correction deserves its own review trail.

Test plan

• Pattern A correctly labeled DESIGN PHASE with TD-OGIT-G-SLOT-1 reference
• All 15 patterns A-O use canonical letters from W1 master
• Shipped patterns (H, I, N, O) cite specific files
• Substrate files (world_model.rs, persona.rs, tokenizer_registry.rs, etc.) preserved in appendix
• Pre-work checklist directs design-phase work through TD-X rows
• Append-only governance preserved (this file is not in the append-only set; pure overwrite is the correct mode for a mandatory-read entry-point doc)

🧠 Future agents who read tier-0 first will now correctly route DESIGN-PHASE work through the TD entries instead of skipping it.

🤖 Generated by Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: dff65796b5

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]

Keep Pattern I marked as open until it is wired

When future Pattern I work is triaged from this table, marking it as shipped will cause the ractor/background flush integration to be skipped even though only the pure CycleAccumulator helper exists: repo-wide rg CycleAccumulator only finds the contract module and its tests, and the authoritative master plan still lists Pattern I as open with the flush policy unspecified. Please keep this as open/partially shipped until an actual background L1 actor or caller wires the accumulator into continuous cycles.

Useful? React with 👍 / 👎.