There isn't really a solution at this time.
OpenVSP must interpolate the airfoils across a blended region. It does the interpolation in a frame where the airfoils have unit chord. So, in that scale, the t/c of the root/tip airfoils of that blended section are very different.
OpenVSP basically creates intermediate airfoils with linearly varying t/c of the blunt trailing edge. When those get lofted, the chord varies in a nonlinear way -- and so your dimensional thickness of the blunt te varies as it does.
Rob