Times up!) I figured it out. The statement that composed the turtle rotation with the new transform needed to be separated into a couple of steps to isolate the composition of the rotations from the translation.
Making this cannon: method simulation was a fun exercise mainly because the screw-ups were fairly entertaining to watch, and there were several of them. This last one was a pretty mild one by comparison.
Here is a challenge that I don't expect to complete myself. Figure out how to accomplish the same simulation using only the provided turtle movement commands instead of the underlying matrix transforms.
I'm still toying with the idea of holding a turtle script contest, but haven't gotten any prize donations yet. ;) I guess it will have to be virtual MockBucks.