It might not be possible to determine the winding direction of a single part, but it should be easy to determine the difference between two otherwise identical parts with windings in opposite directions (if the winding direction actually impacts performance, that is). I'm guessing the SRF of each winding would be the best indicator.
For this design, telling them to always start winding from the same end of the bobbin, and always spin the bobbin the same direction, should be sufficient. Though you show a few layers having turns pushed towards the ends of the bobbin, not sure if they can actually do that without splitting the layer into two separate winding operations.