In short: ignore that remark. It was incorrect. If you want to read the full explanation of my train of thought, read on.

Have a look at this picture, of the rotor in its initial condition (at 0 deg). Notice that the flux of the top (12 o'clock position) magnet is divided over 2 teeth. Also notice the color of the two teeth, which is an indication of flux density.

Now, look at this picture as the rotor has rotated 22 deg clockwise. Notice that there now is only one tooth with the same colour as in the previous image. That is, only one tooth has the flux density that previously was in two teeth. Hence my initial idea that that was an illustration of cogging. But it was wrong.

One shouldn't just look at flux density but at total flux, and the direction in which this flux tries to pull the rotor. It can be seen that sometimes it's clockwise (CW) and other times counter-clockwise (CCW).

Both the above situations are stable w.r.t. cogging, as there's neither a CW or CCW torque present. So my initial reasoning was flawed. The image below, however, shows the in-between situation from the above two images (only 11 deg. CW rotation):

In the above image it can be clearly seen that the rotor tries to turn CW so that the magnets 'embrace' most of the stator teeth (or, in physics terms, the rotor and stator try to achieve a situation of least energy). So, in conclusion, there is cogging, but it's not as obvious from the animation as I initially thought.

Actually, as I had just made the animation and was looking at it for a few minutes, I noticed a sort of 'swelling' action, where the red areas were appearing and disappearing rhythmically. Anyway, I think my statement you questioned was premature and I shouldn't have included it in the post. My excuse is that I'm still on the learning curve myself :)

There's only one sure way of telling whether there's cogging and that's to let FEMM give the torque numbers. I've been making another LUA script to make a graph of the cogging torque. At the moment I get the torque figures out but I can't write them to a file. And I don't fancy manually writing down the 90 numbers and typing them in a spreadsheet. The numbers as they appear on screen clearly show that there are CW and CCW moments at various positions of the rotor, so that part at least is working. Just have to try to export the numbers to a file and/or graph.

NSNS
<====> (aluminum strips pushed forward and backward by hand for 5 mins.)
SNSN

The neos I have used are round 19mm diam by 25mm high and the heat produced is almost negligible. I am bundling strips of 25mm x5mm aluminum up to 8 at a time (40mm of aluminum x 25mm)
The resistance felt (cogging) is noticeable but minimal.
The hope is to have cogging that is extreme - so more flux energy converts to heat.

Any suggestions to improve the eddy currents cogging effects would be appreciated.