Which is where the numbers come in. It might “feel” like 200g rotating weight is a lot, but in reality in a proper double blind test the only way any rider would be able to tell the difference is because of the way the bike handles and/or responds to sideways movements. There is zero chance of you being able to detect the difference it makes in acceleration.
Also in reality for riding you have that 2.8W loss all the time (actually a varying amount, IIRC rolling resistance is directly proportional to speed and that 2.8W will probably be at ~20mph, but I CBA checking as it’s not going to make any difference to the conclusion). The maths tells you that the extra weight will only cost you that much for a tiny, tiny fraction of your ride even if you have a very high power output.
So that 2.8W might seem like nothing, but the cost of the extra weight is even more nothing.