Add opt-in PDF page boundary markers

[Michelle Matias (michelleangela)]

Summary

Add opt-in preservation of PDF page boundaries.

• Add the --pdf-page-markers CLI option.
• Add the pdf_page_markers=True Python API option.
• Emit one-based <!-- page N --> markers before every PDF page.
• Preserve markers for empty pages so page indexes do not shift.
• Preserve boundaries across plain-text, table/form, and mixed-content PDFs.
• Use page-scoped extraction fallbacks when markers are enabled.
• Keep default PDF conversion behavior unchanged.
• Add no new dependencies.

Addresses #1317

Example

<!-- page 1 -->

First page content

<!-- page 2 -->

Second page content

Testing

• pre-commit run --all-files 
• PDF-marker, CLI, Content Understanding, PDF memory, PDF table, and MasterFormat tests: 173 passed, 2 skipped
• Full local Windows run: 649 passed, 5 skipped; remaining failures were related to Windows console encoding, Windows file-URI expectations, and unavailable FFmpeg