Something is wrong with the LOS algorithm - it does not account for hindrances thru orchards when the LOS draws through an orchard hex without touching a green dot.
The L4-N5 example is NA, but he's correct in that LOS traced "through" the hex but not touching an orchard doesn't show a hindrance. However, tracing LOS down the hexspine of one of those hexes does. For example, P6-R6 shows a 1 hindrance, down the hexspine of Q6.