Yeah, the margin the script uses to determine if more iterations are needed is kind of tight. Basically the median of the histogram must match the expected value within 1. The idea being that this is as theoretically as good as the script can do, and if say you just left the script going over night and let it do as many iterations as it wanted, you wouldn't want it to stop until it knew it absolutely couldn't do better. What may actually be 'acceptable' to a human is an entire other matter, and probably somewhat looser. I usually just do a few and even if it says more are needed, I just watch it to see if its acceptable and if it is, I don't bother with more.
If you find, however, that the script isn't affecting the exposure value, and that's why it thinks it needs more iterations, then that is a problem.