My mistake. I assumed HFT was involved because the timescale of the "layering" orders and subsequent cancellation (following the bona fide order execution) was on the order of several hundred milliseconds.