There is substantial amount of API changes to allow correct framing of data for aggregation. Query model now has both sets of columns: bottom-up and top-down. Top-Down is new. Upstream code now has context of what data is needed downstream. I also fixed a few bugs along the way, too many to mention