release: v2.0.11 PDF scan vision and subagent wakeup merge

[wicm84266964]

Summary

• Scanned composer PDFs render the first few full pages for vision; text-layer PDFs stay on extracted text. Full visual read requires explicit page range.
• PNG pastes without MIME are classified by extension.
• Background subagent groups merge wakeup policies so a later wakeParent=false cannot disable an existing wakeup.
• Adds @napi-rs/canvas 1.0.9 for page rendering; Windows exe distributions ship the native binary.

Validation

• Local public verify:release: 1,331 unit/integration and 38 browser tests passed, 0 failed, 0 skipped.
• Cross-repository parity: 27 exact paths and 13 reviewed differences.
• Candidate notes: docs/releases/2.0.11-pdf-scan-vision-wakeup_zh.md.

This is the v2.0.11 public release candidate. Tag and GitHub Release follow after required CI on this PR and on main.