I don't think that ReMine made the claim that all the substitutions would have to be single nucleotide substitutions. I believe in his book (my copy is on loan) that he discussed the other types possible (including insertions, deletions, etc.).
Yes, you're right there because just today, I was talking with a YEC'er and he brought up Haldane's Dilemma. Well, I asked about the allele or single nucleotide thingy and he said that Remine for illustrative purposes broke up the maximum 1667 substitutions, into rough groups something like 1500 of them would be point mutations, then 100 another kind of mutation, etc. So I'll amend my opinion: He's not entirely dogmatic about the numbers. But I still think he's willingly choosing to overlook a well known fact about single nucleotide substitutions that I believe has a bearing on his argument.
However, keep in mind that a gene substitution can involve only a single nucleotide difference and ReMine states that according to evolutionists that substitutions in genearl are "typically" single nucleotides, not whole new genes.
Its true that most mutations are point mutations. So it would seem logical to also say that most beneficial mutations are of this sort then. But are they really?
As far as I know, single nucleotide substitutions are mostly neutral so there'd be nothing for selection to work on. There are also some point mutations that are deleterious and even fewer that are beneficial. It doesn't seem like it would be the main source for beneficial mutations. Its not impossible, but there's other info to consider.
I've read that a number of new genes beneficial mutations are found to be duplicates of other genes, modified over time by deletions, insertions, etc. to have a new function. Now that takes multiple nucleotide substitutions. Thats why I don't think Remine can go with even a rough figure of 1500 nucleotides plus any remainder. But all of this is really useless number wrangling, because the Haldane model itself is an unproven theoretical model.