For anyone looking for a solution to a similar problem, this is the workaround I've settled on:
First create additional data points that are 1.1 times the x and y data max subset (just "outside" the original max subset), and fill the corresponding z points with NaN values. Add these x,y,and z pts to the original data sets to be plotted. This will at least prevent the area above the max line from being filled in, though it also makes the fill contour around the max line very choppy. To get around this, add in additional data points along the max line using interpolation until the contour plot looks acceptable.
Cheers!