[Peer Review] Give feedback on potentially bad choices of settings

[cthoyt]

It seems that the default setting of the command line client is PreComp. I tried putting in a network with ~500K nodes / ~ 9M edges and after it had been running for a while, I realized that I had made a poor choice based on your insights outlined in the manuscript. Instead, I should have probably chosen the SparseOTF mode. Since you already make suggestions in the manuscript on what the cutoffs for size and density are to go between each, it would be helpful to also output a warning after the graph has been loaded if the mode does not match the selected mode (especially since selecting the mode is an optional parameter, and people, like me, will try using the default parameters unless otherwise encouraged)