Yes, their method for working with huge volumes of keys is much cleverer than what I've shown (and the GCD is just better outright and I should have used it even in my illustration)
However, I am surprised that you say "They point out that OpenSSL adds a small amount of entropy ...". Do the authors actually claim this? I can't find it in their paper, indeed the paper seems to make out that the origin of the shared factors is unknown to them, and other commentators have put it down to unknown proprietary key generators rather than the easily inspected OpenSSL. Perhaps they have written about it elsewhere?