Default text extraction doesn't get text from annotations

[duckduckgrayduck]

This causes people to run OCR when just pulling the texts from the annotations would make more sense and would be more computationally efficient.

[duckduckgrayduck]

Example document:
https://www.documentcloud.org/documents/25604601-dossier-recu-complet-le-3-mars-2025/