IIRC (have not read through all of this) the back-EMF argument depends upon crossover+cables+amp output impedance being such that intermodulation happens. Superconducting wires and ideal voltage source (0-ohm) outputs would not exhibit any change; real cables and amps will but "it depends" if it is audible. Assuming the shorting strap on the speaker is the lowest-impedance path, and not the cables back to the amp, then there will be intermodulation among drivers through the crossover elements. I have always thought that, since most cabinets do not internally isolate the drivers, this effect would be swamped by the acoustical interaction (e.g. woofer's air movement in the cabinet modulating the midrange and tweeter cones), but I do not know.
As
@RichB said bi-amping does not provide more power, or not much, particularly if it is the "passive" scheme of most AVRs so the same signal is passed to both amplifiers. The load is different for the two amp outputs, but the output voltage is the same, so no voltage headroom and no power increase. (There wil be a little due to differences in current demands but that is usually very slight.) Bi-amping with two 100 W amplifiers still means no driver can ever draw more than 100 W -- it is not the same as using a 200 W amplifier. If you do active bi-amping, splitting the signal before the amplifiers, then you do gain some headroom since the lows and highs are split so a single amp is not handling both.
FWIWFM - Don