Number three. The goal is to eliminate variables of motion. If the booster is against the left arm, then the left arm doesn't need to move, and the booster gets a bit of stabilization from that arm. At the same time, the target position for the right arm would be known precisely.
I'd let the booster settle onto the rails instead of trying to raise the arms. Once they're up to speed on cycling boosters, they can turn the whole booster catch into one fluid motion while the booster is gently descending.
Come to think of it, a cross piece that sticks out at a 90 degree angle from the left arm would also help. The booster could nestle into the crook of the arm and that would make it very stable while closing the right arm. The one drawback to this whole leaning thing is that it produces steel-on-steel contact. If the boosters are going to be cycled many times, that's going to wear the surface of the booster. I suppose they could introduce some kind of crush barrier on the booster and/or a cushion on the arm.
As I recall, the right arm has received all the testing and hardware upgrades, so they may actually be planning to do it this way. Hopefully, we'll see in a month or so.
