Locating the start of the loop range at transport start does not count as "have looped". This was preventing pre-roll when starting to loop, which in turn caused data/event loss because the start/end sample ranges would be incorrect inside process