I guess I misunderstood. I thought the issue was with timing, which I thought would be a particular issue at unload/load. So, I figured that they'd have to have live people getting on and off the trains to establish the right rhythm/flow.
It's a build-up. To unload, load, and dispatch there is a fixed sequence of events that must happen in a specific order. First, the staff need to learn that sequence, and be able to do it reliably, time after time, every time, and complete each step within the time limit required. The more trains there are out, the tighter these time windows are. Staff train by learning the sequences, then increase the # of trains which pushes them harder. Then, they add riders into the mix, start low and build up again. Guests add new delays and variations the staff need to compensate for. This is just the platform training they must go through.
But even before that, you have the ride system testing itself. More on that in a second...
I didn't think that the issue would be trains running into each other or anything. I figured if they aren't "efficient" enough the trains would have a stopping place before unload and people would just be stuck there in the vehicle. Just like how on many rides, we see the vehicles back up before going to unload.
There is the theory of operation, and then there is what happened when it got built and real physics and real constructions are in play, not the modeled stuff. The ride computer which controls the safe operation of the train's brake zones and emergency stops often needs to be dialed in to compensate for the train's real world operation. The ride computer relies on sensors around the track to determine the train's locations, speed, etc. It's not uncommon for these sensor networks to be fickle after initial installation and be the source of lots of problems requiring extensive shakedown testing to get the system reliable. This, combined with with the ride's software often are problematic when trying to get a ride system to run at full capacity. The more trains, the tighter the timings, and more safeties that might accidentially get tripped, etc.
It's not uncommon for a new system to not be able to reliably run at capacity without 'errors'. These errors could be tripping a stop or even e-stop unintentionally due to the ride computer thinking things are worse then they are, or even losing track of a train... or at the extreme, a failure to maintain train separation and getting a collision. But even without a collision, there are lots of errors that can happen with the ride controls.
As to the stopping spot, yes, it's common to have a hold position there... and you would have hold spots elsewhere on the track as well so if there is an e-stop, all trains can be held in their own zone without collisions. But for all this to work, trains must maintain the proper separation.
I'm sure the show scenes make this difficult as well by effectively reducing the size of the track and zones.