First of all, in the context of B&W film, using the green channel (or biasing towards it, when mixing channels from RGB - in order to convert the color image data to grayscale) is pretty much logical, since 50% of pixels in the sensor are green sensitive. You'll need to interpolate from less data when using the other two channels (25% of pixels each...), whereas you have 2x data to interpolate from when using the green channel. BTW, which channel(s) are selected and their mixing ratios (if more than one) will affect the grayscale result very much in case of color film; therefore suggesting to use green channel only may not work each time, depending on image content and intent...
OTOH, I haven't noticed that much sharpness difference / color fringes (if any) with my Epson 2450 - the sample above show serious problems. Maybe it's also (or maybe, "more" - if you like) related to the optics of the camera; I mean the effect may have caused by using a non-achromat or non-apochromat lens??? (Also, was that image scanned single pass or multi-pass? Multi-pass may raise problems if there's play in the positioning belt/system...)