fix(apocalypse): ignore stale PDF search results

[fly1d]

Summary

• make Emergency PDF text search latest-query-wins so delayed extraction and errors cannot overwrite a newer query
• render pages offscreen and commit only after operation ownership checks, preserving the latest page, zoom, and resize request
• invalidate searches on manual navigation and guard repeated-target request rollback

Reproduction

1. Start a search whose text extraction stalls on page 2.
2. Submit a second search that finds page 3 and finishes first.
3. Release the first extraction. Before this fix, it could still navigate to its old match and replace the newer status and canvas.

The regression test also covers pending-page resize, quick next/previous navigation, repeated same-target requests, stale failures, and search-result render failures.

Testing

• node test/run.js (1818 passed, 1 failed; the unrelated failure is the missing upstream artifact dist/webbrain-chrome-32.1.0.zip)
• npm run test:toolbar-guard (33 passed)
• npm run test:security (60/60 passed)
• git diff --check
• Chrome/Firefox source parity checks

[vercel]

@fly1d is attempting to deploy a commit to the esokullu's projects Team on Vercel.

A member of the Team first needs to authorize it.