The photo control won't be damaged by switching the power off, the switching device will take the beating. I've had many PC's that get switched off during the night by timers and the PC's lasted for years.
The switching device takes the beating well, but by that exposes the line from the switch to the ballast by the high voltage overshoot. No problem for the switch, but it could be a problem for the photocell electronic.
And if the timers were using triacs, there was no overshoots - the triac can switch OFF only as a consequence of the current disappearing.
Clearly, the PC's die, when connected behind a switch, but survived when after an automatic timer (you just wrote that), so you can not blame just the photocells or ballasts alone, it is only the combination of all the three elements (switch + PC + ballast).
The photocell contact is designed for all the overvoltages the switching causes, but not the power supply for the control part. Normally that is not any problem, as the control part is connected to the permanently present mains, where would be no overshoots at all. The overvoltages happen behind the switching element, so normally on the output side, where is nothing connected (just the ballast).
Now when you add a mechanical switch upstream of the photo cell, all behind that switch would become exposed to the overvoltages and it is a question, how that would be handled.
With the old fashioned photocells consisting just of a photo sensitive resistor and a relay (either electromagnetic or thermal), there is nothing sensitive to the overvoltages (all the control part is just the photo resistor in series with the relay coil/heater directly on the mains input), so they handle that well (500V spike is just about 4x the rated voltage for a brief moment, so not enough energy to cause any overheat).
With electronic units, the controller needs a low voltage supply (~24V for relay output, anything between 5..24V for the triac versions).
And that could be done either by dropping resistor (that would remain tolerant to the overvoltages, but it delivers only limited current, so can not feed a relay, so limited to triac output)
Or a dropping capacitor. That does not dissipate virtually any heat, so allows for higher current, so able to feed the relay output. But as the capacitors impedance reduces at higher frequencies / fast transients, for the overshoot it becomes nearly a short circuit. So with the same 500V spike example a part designed to operate at about 24V suddenly get the full 500V, what means ~20x overvoltage. And that becomes way too much, even when there is a resistor in series (for exactly this purpose; but not designed to accept too many of such pulses)
As I wrote, without knowing what is damaged, it is not possible to tell for sure, what was the reason.
And even with cutting the electricity off upstream of the photocell, it very strongly depend on where and what cut it out. For an example when the power is cut on the utility side, the voltage just disappears, so no overshoots whatsoever (that is, what the photocells are designed for).
But if you have just a switch upstream without anything else, you easilly end up with high 100's V spike, when the ballast exhibit inductive behaviour. So the photocell could have problems.
When the ballast is of a power factor compensated series choke or HX (so has a capacitor on it's input), there will be no turn OFF overshoot. But there would be some turn ON inrush current, unless the turn ON is timed so, it happens at zero voltage across the switch. So the photocell won't have any problems at switch OFF, but it might have problems at switch ON (in case of a triac as the switching element - it tend to trigger by itself just as the response to the dV/dt from the mechanical switch operation, so out of control of the controller part, which could normally control the timing so to prevent the inrush current).
