But, if you collect yeast in mason jars and shake the jar you should be able to scrape the top layer and get to the middle. Maybe this isn't perfect but it is what I have used in the past with great success. And I usually cold crash on the yeast cake and I can't say I have ever had any problems with this method. That said, I have never gone more than 3 or 4 gens with carboy/bucket yeast harvesting.
Using conicals, it is very nice to be able to select yeast from the middle layer with racking arm.
Mark: If oxidation is not a problem on un filtered beers why is it that on our bottling line at YHB the beers that come off the line mis-capped are set aside and hand capped later and reserved for brewery consumption. And after about 2 weeks there is a definite flavor loss in the mis-capped ones as opposed to the capped ones. I can't understand why this would have any other explanation except that is is oxidation. And this is on a wheat beer with plenty of live yeast in suspension. If having live yeast in suspension was a silver bullet as you assert, you would think that those beers would be just as fresh as the others that are capped on foam a half second later.
All that said, I used to rack my homebrew into unpurged corny kegs before I "knew better" and back then I can't really say I experienced any major problems. I would be really careful on the racking though and purge the head space immediately afterward. If you have a lot of co2 still dissolved in the beer and you rack really careful the co2 coming out of solution should displace the o2 in the head space. But, still, I think it is far better to just purge. After handling hundreds of kegs and thousands of gallons of beer at a working brewing I can truly attest to how important it is to minimize o2 pick up post fermentation at every level. Because, infection aside, oxidation is the number one factor in beer freshness - hands down - even on unfiltered beer. I do believe that yeast in suspension helps to mitigate oxidation, but I do not believe that is is a magic bullet.