fix(wiki): prevent Unicode memory recall from closing MCP transport

[rabii-chaarani]

Unicode near a highlighted search match could put a snippet endpoint inside a UTF-8 character. During wiki_memory_recall, the resulting Rust panic exited the stdio MCP server and surfaced to clients as Transport closed.

Move excerpt endpoints outward to valid character boundaries before slicing. Add search coverage for every interior byte boundary of two-, three-, and four-byte characters on both sides of ASCII and Unicode matches, plus a real MCP process test that records and activates a Unicode memory, recalls text and structured results, and answers a subsequent ping. Record the verified diagnostic lesson in repository memory.

Validation:

• Both new regressions failed before the fix with a UTF-8 boundary panic, including in the MCP server's main thread.
• All 98 k-wiki tests passed.
• cargo fmt --all --check and strict Clippy passed.
• Repository policy gate and optimized executable MCP smoke passed.
• Recommended wiki validation, link checks, and site build passed without diagnostics.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-30T21:57:46.594871Z	5bc3883	PR opened

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.