The contiguous cubies option never worked well and was removed. Try reimplementing it using Roice's suggestion in message 1232:
"There is a way to make the implementation of this much easier, which I used in M120C. Instead of shrinking the stickers based on their centers, you shrink them from appropriate points on the cell boundary. You also make it so that when the cell size is set to maximum, it fits the cell boundary of the parent polytope (which is a very natural choice anyway). Then you will always have contiguous cubies when the cell size is maxed out, regardless of the sticker shrink value. This avoids an awkward dependence of contiguous cubies on both the face shrink and sticker shrink parameters (requiring constraint calculations to get a working set)."