Deduplication can only be done if the type have been printed. We perform that printing only at the junction between the reconstructed identifier enclosings and the ones from the tree nodes because we often want to keep both. All other duplicated ranges are removed.

Since deduplciation has been introduced a while ago the results of type-enclosing (the number of enclosings) have been unstable. This cannot be solved easily on the server-side without printing more types which can lead to performance issues when large modules are involed. We now leave the responsability of deduplication to the clients.