This is a pretty cool idea, and hopefully not too difficult to implement.
You could set 2 different layer thicknesses, one for "steep" regions (thicker), and one for "shallow" regions. Then, allow the user to set the limiting angle.
Once layers associated with shallow angles (as measured from the XY plane), the software could switch to that thinner layer height (at least for the perimeter passes).
This would make for really efficient printing, where you would blast through more vertical areas with thick layers, and crank the resolution up where it counts.