[HPsauce]

Well technically of course it can't see which direction, but it can potentially see "interim" positions and so infer direction.
That's assuming the human hand is making "normal" movements and it remembers what the previous position was.
Remember my gear(shifter/selector) position lines are only estimates, I didn't get that information from an F125 switch connected to the cable.

But given the tracks very obviously aren't exactly aligned/delineated around precise locations of each stick position there has to be some logic to it..... doesn't there?

In particular I note that pin 3 will get a signal as the shifter moves out of Park but well before it arrives in Reverse, though that is of course a longer than average travel.
Whereas pin 2 loses the signal only when the shifter is fairly precisely centred in Reverse. I suspect some subtlety of that is related to your problem.
It actually looks like, from a TCU view, it goes P, half-N, R, N, D etc.

Also, in the non-TCU (presumed) group the starter relay gets a very lax allowance on Park but minimal slack on N.
There is even possibly an interim position where you could be in Reverse (or at least have the reversing lights on) and operate the starter given the size of the contacts?

I've also only just realised that the four "relay" contacts run in grooves so can't really go adrift whereas the five TCU ones are not constrained and a slight distortion could have odd results, though they are mounted in pairs.