Seems like solar tracking should be sensorless and just use date and clock to predict where to keep it pointing. Honestly, seems like trying to build a light sensor and the logic to home in on the sun takes way more energy than simply allowing a scheduler to align as we already know where the sun will be within a picosecond.
This has been suggested before as a theory, but i don't know if anyone has actually implemented such a system on a diy setup to demonstrate if it's actually any better or simpler than sensing the suns position or not.
True you don't need optical sensors, but then you do need to sense/keep track of the actual position of the panels, either by using a datum point and dead reckoning from there, or by position feedback on the tracker.
Also depends on how sunny it is where you are: where i live you can have overcast conditions for a fortnight or more in the winter. A sun tracking system ends up facing south and not moving anywhere for the duration, whereas a system as you describe would be running the motor daily to faithfully point the panels at the eastern horison every morning and trundling across the overcast sky to the western horizon every evening.