Fix for some sample bugs a + b ? c : d does not do what I think it does. That bug mostly masked a thorough logic error in building the table of counts of possible sequences; there was too much cumulativeness. There apparently remain problems