I don't see how incident metering can be used for metering specular reflections of light sources, or for light emitting objects/surfaces for that matter - both very similar situations. There are times when you want detail in such circumstances: backlit snow and aluminium-painted rooftops come to mind immediately, in addition to the sort of subject in noseoil's picture.

The use of an incident meter to determine correct exposure relies on the assumption that the illuminated surfaces in the picture have a reflectance factor that lies within a certain range of values. The reflectance factor in the lobe of specular reflection can exceed that range, and so experience and judgement, or a reflected-light meter, come in handy.