That would be more difficult, expensive, and unreliable. The light is powered by the rotation of the wheel so it already has a reliable braking signal.

How is that difficult, expensive, or unreliable? What could be simpler than a switch that closes when a lever is closed, and turns on a light?

That sounds much simpler, cheaper, and reliable than "detect changes in the frequency of the wheel rotation".

Also, it works like a car's brake lights, which drivers are accustomed to, rather than "start pulsing when you decelerate", which is when you stop pedaling (unless going downhill).

The light is already by necessity hooked up to the wheel for power and adding a switch and wires to the brakes adds another point of failure making the system less reliable, makes installation more difficult because you have to wire up to the hub generator and the brakes, and more expensive because more components.