Query.on('remove') to provide document contents instead of empty data

In my scenario I have a subscribe query which notifies me about inserted and removed documents in a collection. I also have a tree data structure (a filesystem) in which I store these documents (see #145).
The problem is that when the query notifies me of one or more document removals, I'm unable to efficiently update the tree since the documents' data fields (which store information about where the document is stored inside the tree) have already been undefined. The current workaround is, for each removed document, to traverse the whole tree by comparing each stored document with the removed one. Instead it would be nice if the documents' data fields were emptied only after firing the event.

Start by tracing the Query.on('remove') event and the document-removal lifecycle to find where document data fields become undefined. Verify the event receives the document contents before those fields are cleared, and confirm the tree-update scenario can identify each removed document without traversing the whole tree.