fix: improve EPUB image parsing and path resolution

[Aatricks]

This pull request improves the EPUB chapter parsing and image handling logic in the ContentRepository. The main changes make the content extraction more robust, especially for images and nested elements, and improve the way image paths are resolved and matched within EPUB archives.

Parsing and content extraction improvements:

• Replaced the flat iteration over body children with a recursive traverse function to better handle nested elements, ensuring both text and images (including those nested inside other tags) are properly extracted and added as ContentElements. This also improves direct text extraction from non-standard containers.

EPUB image path resolution and matching:

• Enhanced the resolveEpubPath function to normalize relative paths, correctly handling . and .. path segments, preventing issues with image references that use directory traversal.
• Normalized image paths by replacing backslashes with slashes and removing leading slashes when extracting image references from URLs and EPUB entries, ensuring consistent matching regardless of path format. [1] [2]