The root cause of the problem I was seeing is how the slicer resolves the issue of the what to do with the infill when the feature thickness is less than the thickness dictated by the number of perimeter shells.
For example, let's say I'm calling for 3 perimeter shells. Based on my nozzle and filiment, my extrusion width is coming out to 0.6 mm. For a 3 mm wall feature, the solver will try to put in a total of 6 widths of the perimeter shell. In this case, 6x0.6 it will come out to 3.6 mm, which wont fit. The software is smart enough to recognize this and it dutifully drops out the inner most perimeter shell. However, it isn't smart enough to go back and attempt to put an infill in that space either. Thus, in this example, I ended up with 4 x 0.6 of the width occupied by perimeter shells (2.4 mm) and am left with an 0.6 mm gap between the 1.2 mm walls. This 0.6 gap is wide enough that the machine could have made an extra pass and filled it in with infill, but it doesn't. The important thing is this, once the solver starts dropping the "inside" perimeter shells, no combination of infill settings will cause it to fill in the gap that is left.
What you are alluding to is tweaking the extrusion settings to either make the extrusion a little bit wider such that it fills in the 3 mm exactly or make it thinner so that the number of perimeter shells fit in the space provided for in the design file. Both of these things could be done too.