The purpose of my photo was to show you how the wire form/trip wire should be bent. The way yours is bent will cause other issues. If a ball goes past the switch, but does not go up onto the playfield, it will roll back down and get stuck behind the wire form and the CPU won't know it rolled back down. I know this roll back issue is unrelated to the problem you are having. But, as has been mentioned, it sounds like the CPU is not seeing the ball pass by the switch and therefore, tries multiple times to re-kick the ball to clear it from the subway.
My playfield, while it sits in a Centaur 1, was out of a Centaur 2 so that would explain the difference in wire numbers and why my diode is mounted directly to the switch instead of on some solder eyelets on the playfield. At least the wire colors are the same.
Have you checked to see how "long" the micro switch can click closed if you slide a ball up the subway rail by hand? Your switch might be clicking closed for too short a time for the CPU to see it (one of the reasons for disc caps on things like Thumper Bumper and spot target switches). I wonder if adding a 0.05 (or 0.047) ufd disc cap (any voltage is fine) across the switch will help?
Have you tried to close the switch by hand immediately after the game energizes the shooter coil to kick the ball to the playfield? If so, does it only kick once if you do? I know you have said the switch responds as it should in switch test. So we know the switch is wired correctly, and, for the most part, works when closed. It just might be a timing issue.