[mjd]

Thanks very much for the followup article! I suspected that sksksnrpny was correct, and that the direct forward construction given in the paper they referred to would be simpler than the ad-hoc backtracking search I did. But I needed to work through an example to be sure, and now I don't have to because you've done it for me.

Also I'm delighted that you remembered that Matthew P. Wiener thing from 30 years ago.