Look at the facing direction on the victory square for this move. I'm not sure if this is a bug, or working as intended but it looks wrong to me. It's in the opposite direction the other squares on that path; and the opposite direction to where I'm moving. There are no obstacles which would make the character take an indirect path, so it seems wrong for it to face that way.
Facing on long moves has always been a little quirky. I'd like to see an option to select your final facing if you ctrl-click the destination.
I think I've seen the devs mention that choosing the final facing direction was their original intention, but they decided against it because it slowed the game down too much. i.e. Choosing the facing direction is sometimes important, and so you'd have to think about it on every move, and thus you'd spend some time choosing the direction on every move... So games would take longer, but the depth of strategy wouldn't be significantly increased. I'm not really sure what's best for the game, but I'm happy enough not choosing directions - as long as the automatic directions are intuitive and natural! I haven't seen the code for this game, but my guess is that the problem in my OP could fixed pretty easily. I'd fix it by tweaking the pathfinder so that cardinal steps a very slightly lower cost compared to diagonal steps. With that, the pathfinder will prefer straight paths, rather than paths that go diagonal outwards and then diagonal back in again (which is what I presume has happened in my screenshot).
Since this seems to show up on long moves, I thought I'd see what happens with a REALLY long move: That's not the maximum possible, but it's close (range 18). For solutions, karadoc's weighting idea is good. Another possibility would be to look at the last step of the move, and if it was diagonal choose a facing away from the the starting position (if possible).