DKE is not simulating a Georgia runoff because
there aren't enough pollsters putting out samples that include both a Nunn/Perdue/Swafford permutation and a permutation that's Nunn/Perdue alone. None of the five [now 6] most recent polls, for instance, have that..
DKE's current mashup of polls has Nunn 48.6%, Perdue 47.0% and Swanson (L) 4.4%.

29% divided by 40%, results in a 72.5% chance of Nunn winning if there's an outright winner.*
*If you assume Perdue held steady at 13%, then there's a 69% chance that Nunn's the winner IF there's an outright winner.