Hopefully I can derive a signal by means of the magnetic field as suggested. That would be the easiest.
Otherwise I was thinking of using an LED as the sensor as they generate tiny amounts of power when light shines on them. These fans are unlike most others, they are made by ebm and are very quiet. They are side blowing fans. So their turning would only be visible from a few angles. They are very precisely balanced so whatever was attached to them would have to be very light. A very small square of adhesive coated mylar tape, I am thinking might work best, either reflective or white, depending on where the two LEDs were put. So that the LED would last a long time, I am thinking it would be best to run it at a lower power than its rating.
The amounts of power generated by an LED is very small. When I have connected my voltmeters to LEDs under the lights I currently have the values keep changing. This likely is because of my LED lights flickering at 120 or 60 Hz (I'm in the US) . Would need to look at the 'output' under the sun with an oscilloscope or similar (and shielded cables) to find out what happens under a steady source of light.