I still find it amazing that I genuinely enjoyed the university course that basically amounted to “get into teams and design and build a better heatsink than the rest of the class”.
One thing to emerge from that experience is the value of tailoring your heatsink to the actual operating conditions. Very approximately, you want to maximize convection for things that are running for a long time (such that they heat-saturate the adjacent material, and achieve something approaching a steady state), and maximize conduction for quick dissipation of heat (at least until you run out of heat capacity or temperature differential). Radiation doesn’t really matter (because the temperature differential and surface area relative to the amount of heat are both fairly small).
With that in mind (and not having done the calculations), I think active cooling (i.e. forced convection) is probably a better choice than a small passive heatsink. There’s a lot of thermal resistance between the heatsink and the flux yoke of the motor, and yet more between the flux yoke and the case. As a result, you might find that the little fins don’t even get hot enough through conduction to drive a decent thermal differential. Even then, you’re more interested in the armature staying cool than the case. The best way to ensure that is with moving air flowing through the motor.1
Failing that, an actual heatsink (to store heat rather than convect it away—kind of like a CIM motor) would probably keep the motor cooler for short durations. They’re kind of heavy, however. (I haven’t done the math to see how heavy this would be, for a given performance and a given duration.)
1 Someone, maybe AndyMark or BaneBots, needs to run a nice little controlled test: compare performance of motors with various levels of ventilation and heat-sinking.