Is it better to compare to the chamber, or compare to a once fired case to find the .002" of bump?
I don't think it matter which you pick, it's the act of picking one that's repeatable, which is probably going to be the once fired since you can mark one and keep it forever. It's going to be super close to the chamber anyway in length.
to the OP, here's my take so far: I actually got all my brass once fired before I headspaced my prefit. Then I used my go gauge and the once fired brass to get as close a fit as possible to the previous guy's gun, which meant I set it around go+.003. Then I try to size right around .002-.003 shy of that original case, which is now functionally identical to a case fired from my gun. And you can tell when closing the bolt if you're even one or two thousandths off, longer or shorter. .002 seems to be an amount you can just not quite feel in the bolt, so I feel like that's part of why it was picked.
So I do all that prep work that makes me feel like I know something, and then I go and lube unevenly, get stuff caked on the press, have die die wander on me, and only get about half my cases within one thou of the target. Some are bumped by .012, .008, .000. Some are longer somehow. Then I get really nervous and waste a lot of time fussing around trying to diagnose and fix.
Then they all shoot into the same .3moa hole without fail.
I think the concept is sound especially at longer ranges, and the idea of working the brass less and generally having fewer variables is certainly good, but the likelihood you're missing something is low. Any FL sized case is going to be MUCH closer to chamber size than new brass or factory ammo. A basic FL size in accordance with the instructions that come with the cheapest die set is going to be a very good fit. Better is always a good thing, but driving yourself bananas isn't.