I don't think the knight should be allowed to move in a full circle. It should be disjointed - four arches. Allowing a knight to move two squares orthogonally kind of makes it a completely different piece.
I wondered about this too. I think I resolved it in my mind by seeing the game as granting every piece one degree of freedom that goes from discrete to continuous: for most of the pieces, it's distance from the origin, since they move in straight lines. But knights are already more constrained than that anyway, so for knights, the continuous DoF is axial instead.
This also makes sense when you put any given piece in the middle of an empty board and plot all of its potential moves— for every piece but the knight, "joining up" their possible landing sites gives continuous distances from the starting point, but if you join up a knight's landing sites, you get a circle of r=√5.
So in short, I agree with the OP's decision on how to handle knights.
Yeah, I basically tried to get as close to the normal moves as possible while also keeping it simple. Also all of the pieces in Analog Chess are so overpowered I think it's ok that the knight is buffed a little bit.
It's not moving 2 squares orthogonally. It's moving sqrt(5) squares orthogonally. In normal chess, the knight can move to any square which is exactly sqrt(5) units away.
This also makes sense when you put any given piece in the middle of an empty board and plot all of its potential moves— for every piece but the knight, "joining up" their possible landing sites gives continuous distances from the starting point, but if you join up a knight's landing sites, you get a circle of r=√5.
So in short, I agree with the OP's decision on how to handle knights.