I said it was probably the seat stay length 10 posts back.
Horizontal dropouts complicate remote diagnosis to some extent. If one seat stay is too long, you can move the wheel back on that side and it makes it a little better at the brake, but the front is off to that same side. So the pictures show what would happen if the drive side stay was too long. Possibly. There is no guarantee that the chain stays aren't catywumpus. If the NDS chainstay is shorter, then it would have a similar effect. Or a little of both, it's not off by much.
I saw a video by a well known builder who attaches the seat stays at the top, checks the alignment of the dropouts vertically and then adjusts them if necessary. On the bike I saw, he had to bend one of the chainstays down a little.