A frame's around 15% of total drag and around 3% of total mass. So I think if you buy this "relative magnitude" argument you're on the wrong forum. Nobody denies we're obsessing with marginal differences here.
I was more interested in handlebar-rider interactions than frame-rider interactions. They include handlebar differences in their test, which they claim is 90 grams (sic) drag reduction. So if I assume bars interact with rider (I think it's no-brainer they do) and matter much less than this, try adding 90 grams to the Trek Madone 7.9 curve here and see if that affects things
Doubtful. This is the forum where people buy certain bike parts to save them a few grams of weight. And spend good coin in that process too. If anything, this is a perfect place as people are looking to save minute amounts of weight to make their bike lighter and "faster". However ignore the benefits of being more aero. I also would not call those marginal differences myself, but we probably have different definitions of "marginal." If I could save watts and get "free speed" I would be all over that. Especially if racing put the food on the table for me (which it doesn't).
The typical number as I stated above is 80% of a rider's power is used to overcome aero drag. 2/3rds of that is rider and 1/3rd is bike. That means the bike makes up around 26.67% of the total drag, rider makes up 53.33%, and 20% is other. I would not consider the frame drag to be marginal as there is time to gain there. 90g of additional drag is a lot. At 40 km/h you save 50s over a 40K for 100g of drag reduction. I will take the benefit of being more aero. Being aero also works riding in the peleton, not just being on a solo breakaway.