Its that time of year when I get frustrated with my 12054 track section - love the fast track otherwise, but losing the ability to activate the dump cars really kills the benefit. I was getting ready to graft my working 027 operating track section in to a section of fast track when I remembered this thread from last year. Switch update looks easy enough. I was then compelled to look at what is inside the dual button housing connected to the fastrack operating track. The series of cantilevers contacts is indeed a bit silly, but I quickly realized that fully pressing the "unload" button did not bridge all of the contacts. The last (lowest) contact was not bridged. If, instead, I pressed on all of the unload contacts from the backside (ensuring they were all touching), the track section now worked flawlessly. So, it does not seem to be an electrical issue, it appears to be mechanical. The button doesn't have enough throw to press all of the cantilever elements together. To solve this, I simply bent the last contact arm up slightly - it was a bit tricky to do this given the flexibility of the arm, but if you wedge a small screwdriver blade under the arm at the bent section, biasing the arm down, you can then push the end of the arm up (at the contact). Its a little finicky to bend it just right to make sure it isn't connecting in its relaxed state, but doable with a bit of patience.
I closed up my housing after bending the arm, and now the track section works perfectly. After I closed it up, I noticed the two small indents on the backside casework plate. I suspect these are supposed to be the end stops for the cantilever arms - helping all of the contacts push together when the button is depressed (but not successfully doing so). I didn't try this, but instead of bending the cantilever arm, I suspect that these indents could be built up on the inside of the plate with tape or thin rubber sheet - effectively moving the stop up closer to the button. A bit of trial and error with added thickness will likely result in the same improved performance as bending the arm.
Hopefully this makes sense, let me know if you have any questions.
Pic attached for reference - last (lowest) cantilevered arm has already been bent/biased upward in this photo (there is a gap when in the relaxed state, tough to tell from this angle).