Comment by emilfihlman

I think there's a "bug" (a bad heuristic?) in the A* code, since it produces easy suboptimal "solutions".

https://emil.fi/m/astarsuboptimal.png