Now you're getting into totally analog matters. First of all, you could do such a thing, but since each diode will be a little bit different, as will every phototransistor, in its response to current, light, and the relationship between them, you'd have to calibrate each one very carefully.

The way in which the fixed disk drive worked was that took a "guess" (a table lookup or calculation) for the DAC voltage that would produce the appropriate excursion of the head stack, and it had a servo track under the servo head, or not (the same strategy works with embedded servo, but not as simply). If the signal from the head/media interaction was sufficiently high, then it accepted that and would proceed to process the signal from the read/write heads. If it was not, then there was an error voltage which would tell the system in which direction to bias the coil current in order to effect a correction. All of this is pretty "digital" in nature, though most of it involved processing of analog voltages. A comparator could tell whether the voltage was within acceptable limits.

In your proposed case, you would rely on a voltage proportional, though probably not linearly so, with the volume of light that arrives at the also not very linear sensor. This can be done, but it's not fast and it's not easy, and it relies on very precise voltage measurements. There's no reason why this can't be done, but I'd guess, and it's just a guess, that the approach you're considering will cost about 100x as much as doing it in some way that approximates the more-or-less digital approach that HDD manufacturers have used. One way in which you could do this is with a slotted mask, the purpose of which would be to define the various positions on the arc that the voice-coil actuator travels. Doing that precisely would not be easy either. If, however, you had a light sensor on the actuator, which moved as would the head stack, and you had a global light source, which provided fairly uniform light on what could easily be a fully exposed strip of, say, 16mm film, with .01 mm scored lines every 0.1 mm, you could then determine where the steps were and count them as the actuator moved about. Your sensor then would either have enough light or not, just as the servo head would have enough signal or not. Having the film with its slots very close to the sensor, and having the sensor masked with a corresponding 0.01 mm slot so that it only "sees" light when two slots coincide, you could conclude where you are and how close to ideally positioned you are.

If you consider, for a moment how this could work, I believe you'll see that it can be made to operate pretty easily, aside from the requirement for high precision in manufacturing, which, of course, would rely on your measurement and fabrication skills.

RE