We produce units in tens of thousands and some of the equipment don't use just _one_ optocoupler in situations where isolation or safety is involved. They are quite affordable.

But you are missing something. You may experiment how much you want with your 8051. The result of an experiment only says something if it kills you or the processor. For situations where you and the processor survives, you still don't have any valid results from the experiment. If I turn on a light bulb once and it survives, I can't use that to compute the expected number of times a light bulb may be turned on. And I can't use that experiment to conclude what happens if the light bulb is hot or cold before you turn on the power. And I can't use the experiment to conclude how vibrations will affect the light bulb. Or under-/overvoltages. Or if it makes a difference if the lamp gets DC or pulsed DC with very high crest factor.

I have optically isolated serial converters, isolation transformers, battery-driven isolated differential scope probes, ... allowing me to access "hot" units. What was your plan for debugging your unit? Do you have a rotary transformer and climate chamber so you can test your 230V units at 300V or more for extended times at elevated temperatures? Can you inject high-voltage peaks on mains?