Speaking of phase-change materials, why not fill the flashlight with them? It still wouldn't be steady-state of course, but I envision the fan eating up all the power. Heat pipes might help, but they still leave the problem of heat rejection. Evaporative cooling pad?
Idea: using PCM you could make a flashlight that you recharge in the oven.
Combining ideas may make sense here. A flashlight with PCM and a black handle that you can leave in the sun, and optionally wrap in a soaked cloth. Perhaps PV makes more sense, but I haven't run the math.
Of course, you'd also want a thermos to carry it in. ;)
I guess it also depends on the relative importance of mass vs. volume optimization.
http://en.wikipedia.org/wiki/Lunar_Roving_Vehicle#Wheels_and...
Basically, the heat was stored in the wax, which melted. Then, radiators would pull the heat from the wax out into space when the Rover was resting, which would solidify the wax again ready for the next run.
What about a series of smaller copper tubes that then stick out the bottom of the handle to transfer heat from inside to outer air?
Think of sink/source temperature, cost/mass/performance/reliability tradeoffs, air vs. vacuum, power requirements, etc.
http://boingboing.net/2011/10/26/coffee-joulies-review-the-e...
Maybe there's some way to lower the melting point of paraffin?
Evaporative cooling could work, but if you are planning on even including 20ml of cooling liquid, you could just as well put a CR2032 battery in there, which'll power the thing just fine.