I don't know the answer, but perhaps it is so that if the longitudinal seam fails in service, if:
1) the seam is toward the top, the pipe will blow downward and is captured by the earth surrounding it, or
2) if the seam is on the bottom, the pipe would be blown into the air, where it lands, no body knows.
Rotating the longitudinal seam slightly between pipe lengths will hopefully attenuate the propagation of the crack failure along the longitudinal seam. Since this failure mode propagates near supersonic speeds, I wonder if it really makes a difference. I suppose it would to a limited extent if one assumes the weld and HAZ to lack some of the notch toughness of the parent pipe material. Usually the failure is arrested by a valve body placed in the path of the failure.
Best regards - Al