docs(agents): adopt the test authoring gate

[code-yeongyu]

Summary

Adopts the test authoring gate from the test audit in a new root AGENTS.md. It covers only the hand-owned tests: the root test/ directory and packages/web. The file also states which paths are synced from oh-my-openagent and must never be hand-edited, and it points the plugin's test rules to their upstream home. The plugin copy of the gate arrives through the normal marketplace sync from oh-my-openagent, not through this PR.

There were no earlier root agent test rules, so nothing needed reconciling.

What the gate says

The four authoring questions, the junk-pattern list and the retention bar, copied word for word from the audit's shared gate text. All eight audited repositories get the same text, with only the heading level and runner names adapted.

Validation

• npm test passes with the new file: 9 of 9 tests. That includes the repository's Hangul, bunx and launch-gating text scans, so the file adds none of that text.
• Documentation only: no code, test or workflow change.

Refs #168

---

Summary by cubic

Adds a root AGENTS.md that applies the test authoring gate to the hand-owned tests. The file marks the synced plugin tree as never hand-edited, lists the test commands for the root, plugin, and web suites, and adopts the four authoring questions, junk-pattern list, and retention bar from the test-audit method.

^{Written for commit 6a708ab. Summary will update on new commits.}

[github-actions]

Thanks for the PR! Changes to LazyCodex land through oh-my-openagent — could you open this against omo-codex over there instead? PRs in this repository can't be merged.

If you're working with a coding agent, prompt it like this:

I opened #169 ("docs(agents): adopt the test authoring gate") against code-yeongyu/lazycodex, but changes there land through code-yeongyu/oh-my-openagent, where the Codex implementation lives under packages/omo-codex. Read my original PR, then open an equivalent PR against code-yeongyu/oh-my-openagent. In the new PR body, link the original PR (#169), explain that it was first opened against lazycodex, and describe in detail what that PR changed and why.