If you haven't already, could you please try the feature and let me know if it addresses the problem? If you've already tried it and it does not address your problem please let me know what shortcomings you noticed or if you think a certain tweak of the feature might address your problems. I designed that feature for cases like this so it would be better to build on top of that feature if possible.

Also it might help you to give feedback by understanding how the feature works. The way it works is it generates a unique token based on the title, url and body of a post. It also has some special logic for cross posts where it strips the formatting and header of a cross post. When a post feed is loaded, posts are processed from first to last, and their unique token is generated along the way. The app also remembers the post associated with a token. This allows the app to differentiate the "original" from the "duplicates". If any post is found to have a non-unique token, those posts are deemed duplicates and are marked as read.