This updates the logic in the path finder for stepping the path once the distance map has been computed. Instead of always choosing the next direction based on a fixed order of all six principal directions, we try to stick to the previous direction as long as possible. This is a greedy approach to minimise the number of turns taken, and thus the number of waypoints and sizes of moves. The resulting paths are still of optimal length.
This does not affect consensus (nor would future tuning of the algorithm), since path finding is no longer part of consensus since #145).
This updates the logic in the path finder for stepping the path once the distance map has been computed. Instead of always choosing the next direction based on a fixed order of all six principal directions, we try to stick to the previous direction as long as possible. This is a greedy approach to minimise the number of turns taken, and thus the number of waypoints and sizes of moves. The resulting paths are still of optimal length.
This does not affect consensus (nor would future tuning of the algorithm), since path finding is no longer part of consensus since #145).