When a space is exported but only a subset of the pages is chosen to be exported, the document tree of the exported PDF still contains all page titles in the space.

This causes generating small multi-page PDF's to require a separate space for each document since otherwise the document will contain unwanted information about other pages in the same space. In my understanding the correct behavior would be to only contain the pages which are being exported in the document tree.

I know the logic for parsing the tree to only include the exported pages in a way that makes sense is very hairy so I'll try my best at later posting a comment with proposal for the logic.