### Next: 88
BUG-60: entries per page in flat view will get how_many from each top-level dir in PATH - causes viewer to get lost in eids for last_eid
BUG-85: once we rebuild data from scratch, need to reset just clean out pa_user_state's
BUG-86: had some face locn data be in { } not [ ] again -- I think we need to stop using locn in the DB and instead use a tmp_locn or something else -- no way to touch the DB data then
BUG-87: using Person->show matches->then view an image shows bug in viewer with msising data in json -- that same image via a filename search works...
	-- seems some faces have an empty locn, but do have a w and h... (ORM - tmp_locn vs locn???)
	# select id, locn, w, h from face where locn is null;
	(count shows 500+ at the moment)

	interestingly, there is a face for todd in this file:
	693, is there more than 1????

	HMMM, or when I delete Richard Tan?? (but 500???)
	still, the face with no locn also is not connected to any file, SO, why is it a part of the view / select data (is it in the set where we do an AI: search for all AI searches, because its more than todd that fails AI)

	OKAY, there is something wrong in the DB on several levels, searched on AI:craig, I get 34211 as a hit, BUT, viewing that file, there is no face at all - feels unlikely, checking DB:

	I deleted all faces with no locn, maybe it is a one-off with all the crap I
	did with the ORM / tmp_locn.... Will see if it comes back, this stays as a
	bug for now