In other words, in what reference system are these collisions (motion superpositions) evaluated?
All three... coordinate space, coordinate time and the scalar dimensions.
That reference system should be universaly applicable to all motion collisions/superpositions (even cosmic motions) because at this stage it is not even astablished whether the observer's motion is material or cosmic.
The way I was able to resolve it (in programming) was to treat the material and cosmic sectors as being 90-degrees out-of-phase with each other, and "zigzag" between scalar and coordinate motion. What I mean is that when you are doing a material comparison of coordinate locations, the temporal aspect of that motion is progressing/gravitating. Once you complete the material comparison, you progress/gravitate the spatial aspect and do a cosmic comparison of temporal, coordinate locations. (On the scalar side, coincidence occurs when there is zero speed between two motions--which is why a molecule stays together.)
It is actually quite an existential dilemma, because you require an observer to create a 3D coordinate system in both space and time in order for the system to "exist." Remove the observer from either the material or cosmic sector, and the entire Universe freezes in its tracks...
Is it possible for two motions to be concurrent in one reference sytem but not in another?
Yes.
Every dogma has its day...