Cogging and eddy current losses aren't the same. In your case you probably still want to minimise cogging (I can't see any reason for maximising them) whilst making eddy current losses as large as possible.

The easiest way to get an idea of what cogging is is by grabbing a stepper motor and turn the shaft. You'll notice that the shaft turns in discrete stepe. That is cogging, the variable torque needed to rotate the shaft, the discrete steps as it turns.

For an idea of what eddy current losses are, take one of your magnets and rub it over a piece of aluminium. You'll notice that it feels as if you're moving a spoon in honey or syrup. That's how the eddy current losses would feel in a generator.

For any significant water heating you probably need much more and larger magnets, with as small an airgap as technically possible. There have been plenty discussions on water heating using magnets in the past. As far as I recall it is technically doable but not the best/easiest way to go about heating water. Have a look at WindstuffEd's experiments too:

http://www.fieldlines.com/story/2007/10/3/195821/664