[codex] Build typed global search results

[tobomobo]

Summary

• Extract global search result types, ranking, transaction lookup state, and app result building into ui-tauri/src/components/kb/search/.
• Add explicit result categories for pages, actions, transactions, wallets, review items, settings, and reports, with privacy tiers and icons.
• Improve txid search UX with exact match, partial match, multiple-match, loading, and no-local-match states; default Enter on an exact txid opens the transaction drawer while arrow-key selection still controls Enter.
• Add local action/entity search for journal processing, wallet setup, BTCPay setup, report export, logs, passphrase settings, settings sections, wallet labels, and safe overview transaction metadata.

Validation

• pnpm --dir ui-tauri exec vitest run src/components/kb/search/ranking.test.ts src/lib/transactionLookup.test.ts
• pnpm --dir ui-tauri run typecheck
• Browser check on http://127.0.0.1:5176: exact txid Enter opens /transactions?tx=tx1 with the detail drawer, no-match txid state renders, BTCPay search opens the setup dialog, and arrow-key Enter follows the highlighted result.
• Independent reviewer re-check: no remaining P1/P2 blockers.
• ./scripts/quality-gate.sh

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 95c15d8b53

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]

Match multi-term queries across all searchable fields

The new ranking logic only accepts a result when all query terms are found inside a single field (title, subtitle, one keyword, etc.), because matchTextField evaluates one field at a time and bestTextMatch never combines evidence across fields. This regresses searches where terms are intentionally split across fields (for example, "home dashboard" for Overview, where both are separate keywords), so valid results are now dropped even though all terms exist on that result.

Useful? React with 👍 / 👎.