The risk in inversely proportional to your knowledge of electronics and the AVR microprocessor chip.
True; but trusting the controller to work as expected is yet another topic imho
. And I tend not to trust the software that much in this case since I neither read the bootloader-code nor did I read the whole datasheet...
You should never use a multimeter to measure resistance if power is applied to a circuit.
Sure, but if I remove power from the controller, I'll probably get even worse results... And the resulting - distorted - value should only be lower than the real one, shouldn't it? So if I get a very high resistance-value, shouldn't there be a pretty good chance that the input-resistance really is high? (Sure, there are cases where this won't apply, especially when it comes to AC/alternating signals in common...)
Edit: Oh, and of course, when the pin is set as output, the measure will also be "a bit crazy"