No, that doesn’t show ATP synthase to be perfectly fit.
That should be easy to test using ancestor reconstruction. If living organisms were perfect in the past, the reconstructed ancestral states should exhibit higher fitness basically across the board. Do you have ANY evidence that this result obtains?