The stator magnets are mounted by a single trimmed 4-40 SHCS and a little washer to make sure only the inner bearing race is contacting the screw or the baseplate.
The rotor magnets are press-fit in place.
The stator mags are mounted coaxially with and above their bearings. The bearings are screwed down to the baseplate with a tiny spacer for clearance. The magnets are diametrically polarized.
These that I'm using now are for 3.5 mm shaft size, in a paired housing that slips into the rotor's central bore, extending down to a shock-mounted base bearing unit.

What I had in there before were just some generic flanged 1/4 inch bore unshielded bearings, and the shaft was a 1/4-20 brass screw. Not the most accurate arrangement.
8 Bearings bought from Stewart-Warner.
The dampers are 6061-T6 aluminum alloy.
The rotor is made of HDPE which is a bitch to machine, and cuts usually come out a little small. I used a 0.250 2-fluted end mill, one pass in width, to cut the slots for the 1/4 in nominal OD rotor magnets, and they press into the slots perfectly--that is, a medium press fit. So if you specify tolerances of +/- 0.0005, which is what I worked to on this project, all should be well.
The stator magnet/bearing holders are Delrin, and it cuts small too, but not as much as the HDPE. I cut for exactly 0.5000, and stopped when the mike read 0.4997. I wanted the magnets to be pretty tight so they wouldn't slip, anticipating the latch requirement.
With the Sherlines, I can always achieve +/- 0.0005 accuracy--they have digital readouts and are very accurate. If I am careful I can get down to 0.0002 in most materials.

Misc Note:

I have little data on variants. If the system works, as I suspect it will, with only the one stator magnet, you might try that first to see if you can reproduce the antigearwise motion, as a start.

The dampers are tuned by trial and error. If too close they retard the magnet too much and it won't synch. Too far and there's no benefit. Once the distance is found, rotating a little at a time and trying the spin until I found a "sweet spot" is how I did it. The effect of the dampers is small, in any case.