I'm not so dedicated. I fill my kegs probably about half way with sanitizer and push that out. the kegs often sit after that so I figure the CO2 mostly settles to the bottom. then I rack out of an open primary through the out post on the keg and leave the PRV open to let the air and CO2 out of the keg. when I'm done transferring I try to evacuate the head space of air as much as possible by replacing it with CO2.
I know this is nitpicking, but over time, the CO2 will not settle, it will mix evenly with the other gases. When the gases first come in contact with laminar flow, they can layer, but they will evenly mix given enough time, and if there is turbulent flow, that mixing will happen much sooner.
This is also the understanding behind that table posted above and why it takes so many cycles to fully purge out the O2. You are mixing it at a specific pressure, purging out a percentage of it, but there is still O2 left in the headspace, mixed with the CO2 that was added. You continue to do this to remove a higher and higher percentage of O2, but it continues to mix in this fashion. This is also why it takes fewer cycles at higher pressure. If the CO2 settled to the bottom, you should be able to just purge once, and completely remove the O2. I hope that makes sense!