drag doesnt grow. if it takes - say , a fraction of a newton / the gentlest of fingertip pressure to get the cranks spinning , then thats all thats required to overcome. the difference between an unloaded (no chain) smooth setup and your draggy setup is that... its not going to multiply when connected to the drivetrain and going down the road at speeds.
That's not true, though your overall point is valid. For a given load, bearing drag increases linearly with speed. The bearing drag from your hubs on a 40-mph descent is twice as much as the hub drag when you're rolling along at 20 mph on the flat.
thats why ceramic bearings, or ooing and aahing over freely spinning hubs is silly. it doesnt really matter.
You're right that ceramic bearings and hubs that spin easily on the workbench don't mean much, but it's not for the reason you stated. Bearing drag is a very small percentage of overall drag; it's almost nothing. Two times almost nothing is still almost nothing.
Also--and this is a bigger deal than many realize--bearing drag on the workbench is significantly less than it is on the bike. That's because bearing drag is not just a function of RPM, but also a function of load. Spinning a wheel in a truing stand doesn't tell you very much about the real-world bearing drag because the bearing is unloaded. You can get a feel for the drag of the seals and grease, but not so much the rolling elements of the bearing.
Everyone loves the feel of old Campy cup-and-cone hubs. They feel spinny and low-drag when you turn the axles with your fingers or spin the wheel in your hands, so people think that translates to low drag on the road. It' doesn't. One drawback of cup-and-cone bearings is that the bearings take a load that's primarily radial, but the bearings themselves contact the cups/cones at 45 degrees to the load. If a cartridge-bearing front hub takes 60 pounds of load, each of the two bearings deals with a vertical load of 30 pounds. But on a Campy cup-and-cone front hub, that 45-degree contact angle means that, to support a 60-pound load, each set of bearings takes a load of 42 pounds, which is about 1.4 times as much as the cartridge bearing hub. As a result, the cup-and-cone hub has about 1.4 times the bearing drag of an equivalent cartridge bearing hub.
In reality, it's not quite so simple...cup-and-cone bearings often have lower seal drag than cartridge bearings, so that helps a bit (a tiny bit). But the upshot is that the drag of an unloaded bearing is a pretty terrible predictor of how much drag that bearing creates under load.
your crank registers torque when it starts to deform from resistance of the chain pulling it back. the draggy bb adds negligible resistance to the system.
its effect on measured power and speed is virtually 0.
Well, with a crank-spider-based power meter, the effect of bottom bracket bearing drag on its reading isn't virtually zero; it's exactly zero. The bottom bracket bearings are "downstream" of the strain gages, and they really don't care whether the resistance they measure comes from mass*gravity, wind drag or bearing drag. But you're right: the difference in drag between any two bearings is so small that it's within the margin of error of any power meter on the market. You literally can't measure the difference with a power meter.
Again, your overall point is right: bearing drag is very low, and bearings don't have a measurable effect on power readings whether the meter is upstream or downstream of the BB. But it's not true that "[bearing] drag doesn't grow." It does, but it increases from a really tiny amount to a tiny amount.