I believe it is simply a matter of programming and firmware.
Unlike RADAR which works on Doppler shifting which can reflect a stronger signal from the following vehicle (although there is firmware to address that issue to some extent), LIDAR works on TOF principles (time of flight), in this case, the filtering algorithms typically pick up reflection of the primary target and count in some regular fashion.
I believe of some reflections get picked up from a trailing vehicle, the time to flight sequence would become unusual and irregular and those with the better and more refined algorithms, get discarded.
This is not to say, that there would never be a circumstance that could create an erroneous reading.
Years ago, during earlier versions of LIDAR guns, such occurrences were likely more frequent or potentially so.
For that reason, a case was litigated in NJ that restricted use of LIDAR (at least in that state) to a distance fewer than 1000 feet.
I routinely have been clocked at much higher distances, though, so that ruling's effects are dubious.
A proper speed clock, should have a tracking history, which should minimize such scenarios.
Hope this is a sufficient explanation addressing your concerns.