what im looking at is the parts of the graph that are above and below the line. as they cancel each other out.
with the exception of the larg spike at the begining, its perfectly symmetrical. same under as over.
now with that large spike, you have half going up, half going down, so that too cancels each other out.

if you look at this set-up under a gauss-viewer, you can see the flux-density as they aproach one another.
areas of denser flux lines equate to an opposing force.

it appears that the outward force on the sliding magnet (in bottom position) is greater than the 'reverse' force on the rotating magnet - this is only the case for the first few mm, as it steeply drops off as the sliding magnet leaves bottom position, and the reversed force becomes dominant.

the lines in the gauss-viewer: i call Fluxii - these lines are equivalent to some # of actual lines of flux.
i dont know exactly how many lines of flux are required to make 1 line on my gauss viewer, but its proportional, so we can assume that each of the Fluxii are equivalent.

now, counting these lines, during the magnetic interaction - we have 8 and 12 ( 12 being the sliding magnet in bottom position). as the sliding magnet moves away and the two come closer to the parallel, the proportion is more 8:8
then 8:4, 8:2, and away from the repulsion zone. after they pass one another, the sliding magnet is too far away to affect the flux lines from the rotating magnet. overall we have a 4:3 ratio of flux-force working against us. Whatever ammount of rotation the system has going into the magnetic interaction phase- is magnetically decreased by approx. 1/4 (in my setup),  resulting in the eventual decay of the rotation.