Early on, we've decided that both approaches are not for us. We'll need something a lot more rigid and sturdy. Eventually, we've settled on a press-fit housing. It's located in azimuth by the pin that goes into a slot in the armtube. There is a screw too, serving as a second locating point and a check on correct positioning of the isolator inside the housing.
Yes, this means that azimuth adjustment
by rotating the connector in the armtube is no longer possible. Don't think this loss should be lamented. This approach is just too crude, and I simply can not accept the resulting loss of rigidity.