If your initialising routines produce any unsafe states, you have coded them wrong.

yes, but the overlooked fact is .. that reset produce unsafe states, thus you have designed the hardware wrong.

The frequency of Is there a sure shot way of ensuring that the out puts are suppressed at power-on and similar posts show a very sad state of the reading of datasheets before designing.

I have seen designs that relied on "reset is too brief to cause any danger/problem/... when the ports are high" and I have (not in combination) seen power supply noise due to e.g. a cap gone bad resulting in enough noise to almost keep the uC in permanent reset while the rest of the circuitry was not sensitive to the spikes. I wonder when the combination of relying on the briefness of reset and PS noise will do harm.

Erik