In case anyone else stumbles across this - my solution has been to use the h3 framework developed by uber and simply bin the polylines in to indexed hexagons with count enumerated every time a polyline passes through the hexagon. Seems to work pretty efficiently