Document formats: .epub, .rtf and .odt

[Phantom-VK]

Problem

Three file formats that matter to the audience, currently refused:

• .epub -> e-books
• .rtf -> legacy Word documents, still common in legal/publishing
• .odt -> OpenDocument (LibreOffice), used by privacy-minded audiences

Implementation pattern

Each format follows the existing _read_pdf / _read_docx pattern in parsing.py. Three separate PRs, one per format.

.odt priority

ODT matters most for the privacy audience the tool already targets, who are disproportionately LibreOffice users.

Libraries

• epub: ebooklib
• rtf: text extraction via simple parsing (simple format)
• odt: odfpy

Scope

Each is a single file handler + tests.
If takes more than 4 file changes, create proper sub issues or branche.