I must agree, a probing accident could definitely result in an unpleasant situation. However, if the current limiting resistor is strategically placed (i.e., not immediately adjacent to the optoisolator), that should keep the voltage at the inputs of the optoisolator pretty reasonable
Not necessarily relative to ground in case the uC circuit is grounded. If the uC circuit float, all things can happen, though not necessarily POOF.

Again, no electronic reason.

Another reason is that it will, I believe, be difficult to get UL approval of the optoisolator circuit. The "easy" way to get UL approval would be to use a UL approved wall-wart with 3-5V Dc output as the "sensor".

I may be overly cauteous about isolating the 220, but a company I once worked for lost a FE eventhough they used "acceptable" design practices. I would not worry if the involved were "regular" electricians, but the low voltages we customarily work with is not conductive to developing "one hand in the pocket" as a normal way of working.

Erik