Edit of otherwise dud message
The removal of duplicates from title lists is a fundamental requirement of most filter combinations if you stop to think about it, say I want a list of items tagged a and B then I do not want to see those tagged both a and b twice.
they are after all considered title lists, even although we do try and redeploy filters in many ways.
I agree we need such a features because the logic of the world does not always comply, however can you try a slightly different architecture to the one you are using?,
Perhaps instead of a field treated as titles use a data tiddler and have a index to each of the elements, in this case element 2 will always be element 2.
Data tiddlers are more like a database and filters spend a lot of time being about titles, I would be concerned if a filter feature resulted in making such title handling more complex. In your use case I thing data tiddlers would be a more robust representation of what you are doing anyway.
The "PR ( #2963) to add that capability. It was ready two releases ago." that Mark mentions may be held back because of this fundamental nature.
Regards
Tony