PDF future architecture and performance ideas

[Amm1rr]

Context

The PDF backlog contains future-facing work that is not ready to implement as a normal Issue:

• keep or remove the currently unused GeometrySyncEngine / CanonicalAnchor infrastructure;
• investigate parallelizing buildPageMetrics();
• investigate virtualized PdfPageView mounting for very large PDFs;
• investigate the incremental usePdfTextFitter loop.

The scroll ownership, IntersectionObserver ownership, and zoom metric rebuild items are already tracked separately as actionable Issues.

Decision to make

Which of these future architecture/performance ideas are worth pursuing, and under what measurable need?

Migrated from the former PDF backlog.