t27_examples: regenerate examples.wasm from the spec that imports find and has

[gHashTag]

Closes gHashTag/t27#10166

Part of the epic gHashTag/t27#10066. The spec change is gHashTag/t27#10169 (the spec itself landed in gHashTag/t27#10157).

What

specs/t27mcp/examples.t27 kept its own copy of find and has, which are also in check_hints.t27. The duplicate-bodies gate of gHashTag/t27 rejects that. The spec now imports them with use t27mcp::check_hints;, and this PR carries the regenerated artifacts:

• wasm/examples.wasm: same size (914 bytes), new bytes. t27c gen-c inlines the imported bodies, so it is still one C file.
• wasm/examples_table.mjs: NO_POS moves to the end of the declaration order; no value changes.
• wasm/BUILD: notes the use and carries the new sha256.

No hand-written JavaScript changes.

Evidence

• Railway lab, t27c 0.5.2, zig 0.16.0: t27c test-report specs/t27mcp/examples.t27 gives 8 of 8 pass and 0 vacuous; with one row widened past 40 lines, 7 pass and 1 fails (negative control).
• python3 tools/dupe_scan.py on specs/t27mcp: examples.t27 uses find and has from check_hints.t27 t27#10169 reports no new duplicate body.
• The new wasm and the old one agree on 748 cases: topic_code over 676 two-word questions and pick over 72 (code, k) pairs.
• npm run smoke: all checks pass.

🤖 Generated with Claude Code

[tri-claw]

Summary by t27-bees

The spec examples.t27 now imports find and has from check_hints.t27 via use t27mcp::check_hints; to satisfy the duplicate-bodies gate. The PR regenerates the WASM and JS artifacts and updates their hashes in the build file.

Worth a look

• wasm/examples.wasm — Binary artifact regenerated with new bytes (same 914-byte size). Any consumer that verifies exact byte identity of the WASM will break.
• wasm/examples_table.mjs:39–43 — NO_POS constant moved from its original position to the end of the declaration order (__DECL_ORDER__). The comment states "which the spec's own laws depend on"; if any tooling or tests assert a specific declaration order, they will fail.
• wasm/examples_table.mjs:15–16 — NO_POS export statement moved later in the file (after MAX_LINES/ROWS instead of right after NO_ROW). Harmless for JS semantics but may affect source-order-sensitive tooling.
• wasm/BUILD:114–116 — Updated sha256 hashes for all three artifacts. Expected for a regeneration, but verify the new hashes match what your build actually produces.

---

_{Read: 3 files, +9 -7, the whole diff. Head 1b86921. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.}

[gHashTag]

Checked each point; no change needed.

• wasm/BUILD hashes: shasum -a 256 of the committed examples.wasm (1f6364dc...) and examples_table.mjs (26462aa3...) match lines 117-118, and the spec on specs/t27mcp: examples.t27 uses find and has from check_hints.t27 t27#10169 hashes to cf6d4d62... (line 116).
• NO_POS order in examples_table.mjs: the only consumer is examples.mjs, which reads the table by name (import * as S). Nothing in this repo reads __DECL_ORDER__ of this table. The order changed because t27c gen-js now emits the imported const after the module's own declarations.
• New wasm bytes: nothing pins the wasm by bytes except wasm/BUILD, which this PR updates. Behaviour is unchanged: old and new wasm agree on 748 topic_code and pick cases, and npm run smoke passes.