This is yet another step where I assume some knowledge of milling machine techniques.
I used a 1/8 inch end mill to machine 70 mil deep channels across the length of the housing. This provides heat sinking and more than doubles the thermally emissive surface area of the housing. The channels were cut front to back with respect to direction of travel of the bike. This allows the airflow from riding to dissipate the heat better. Cross cutting in the other direction resulting in a checker board pattern would increase turbulence and potentially increase cooling effectiveness. I have found though that after my 15 minute commute on cool evenings that the housing is just barely warm to the touch.
There are some excellent comments below regarding the coating of the housing. It turn out that for an application where convective airflow is the dominant cooling mechanism (as is the case here) that coating the surface of the aluminum will not measureably improve its heatsinking performance. In any case I decided to powder coat mine for durability and aesthetics. As stated above, the cooling performance of the housing is sufficient, so overall it seems to be working fine.