When I raised those points about testability and the supernatural, I was not trying to substitute my own reasoning for that of the judge. I was attempting to explain why there is a "centuries-old ground rule" against invoking the supernatural in science. To put it as plainly and simply as possible, nobody has been able to conduct an empirical test for the hypothesis that "God did it," nobody has been able to design an empirical test for the hypothesis that "God did it," and nobody has been able to conceive of an empirical test for the hypothesis that "God did it."
But ID is not about trying to show a supernatural cause or that "God did it." ID is about trying to find patterns that signal an origin through intelligent causation. According to current expressions of ID, irreducible complexity signals design and complex specified information signals design (the validity of this inference is not relevant to my argument here). Just because Behe or Dembski may take the additional step and attribute design to God does not mean this additional step is entailed in the design inference.