Your method seems well thought out and you shouldn't be selecting for flocculation.
I have heard on one of the podcasts out there that yeast companies don't always use pure strains with their homebrewer products. They want to put out a product that has the character they claim, but is also 'foolproof' for the average Joe homebrewer. They accomplish this by using a blend of a hearty strain, like 1056 or 001, with the desire specialty strain, like the pacman. This way, no matter how finicky the specialty strain, the beer will work at nearly any temperature, will generally attenuate fully and will represent the company as producing reliable products. However, this may just be conjecture, as I have not confirmed this with any company myself.
What you may be observing is the elimination of the secondary strain. The pacman strain may be outcompeting the 1056 and the ferment is taking on the characteristics of just the pacman strain. If initially the pac contained 90% pacman and 10% 1056, then the pacman could easily out-compete the 1056 in 4-6 generations.