The start length/C and end length/C are the chordwise measure of your surface (not spanwise). So, it is hard for me to imagine why you would want to set them to 0,1. Start with 0.2,0.2 (a typical control surface) and modify from there, watching what changes as you make changes. For an all-moving surface, I would expect it to be represented by 1,1 for those settings.
However, as you say, that is non ideal. Instead, I would not use the SubSurface capability and would instead just move the entire control surface.
As you point out, this means you can't use the easy control surface derivative code in VSPAERO, but all you have to do is run two cases -- one neutral, one deflected -- and subtract to estimate those derivatives. That is exactly what VSPAERO does.
Rob