track model core and edge changes between each iteration

[nickvandewiele]

I think maybe one lesson from this is that the travis diff models may need to check 1) time for converge b) model core and edge changes between each iteration, extracted from the RMG.log into some sort of plot or analysis to show that we don't stall like this again.