We thank F. Wagner for intensive discussions. Also D. Stauffer and M. Droz have significantly helped in defining good computational criteria for the detection of multiple phases. P. Wagner, C. Daganzo, and M. Cassidy have helped indirectly (and maybe inadvertently), via the discussion of a separate but related paper. Several weeks of computing time on 192-CPU cluster Xibalba and on 44-CPU cluster Linneus was used for some of the computational results.