fix: fetch every attachment of a page when downloading images

[blaipr]

Problem

download_page fetched a page's attachments with a single get_attachments_from_content(page_id, limit=250) call. Confluence Cloud caps that endpoint's page size well below 250 and signals the remainder only through _links.next, which was never followed. On a page with more attachments than one server page, every image referencing an attachment past the cap was reported as attachment not found on page and skipped, leaving a dangling image link in the Markdown.

Change

A new get_page_attachments(client, page_id) fetches the list page by page, mirroring get_child_pages: it follows _links.next when the response is a dict (and pages by count when an older client returns a plain list), advancing start by the number of results actually returned so a server-applied smaller page size is handled. download_page builds its title lookup from the full list.

Compatibility

Download only. Pages with few attachments behave as before (one request); pages with many now download all their referenced images.

Checklist

• ruff check src tests, ruff format --check src tests and python -m pytest tests/ pass
• Tests added for the change (tests/test_attachments.py)
• Docs updated — not needed, no user-facing behaviour described in the docs changes
• docs/conversion.md — not needed, no equivalence changed
• CHANGELOG.md updated under Unreleased