In my words, the above suggestion says:
"If you don't understand how to calculate bias current induced errors, use only FET OPs". I mean to remember bad jokes with a similar payoff. I think, it's still good to have these problems in text books, and of course, the question can be answered. By the way, in some applications (e.g. high speed) we have OPs with uA input currents that can't be neglected.
Regarding the R2*In term in particular, I want to suggest a simple explanation. The equation is based on an ideal OP, except for the input currents, it has infinite gain. So the feedback will zero the input voltage by setting a respective output voltage. With zero input voltage, no current flows through R1, all must go to R2. So the part of the ouput voltage caused by In will be R2*In. And about polarity, for the In direction drawn in the circuit R2*In will be positive. I think it's obvious, that a positive output voltage will drive a positive current into the input.