Minimization results differ for different numbers of CPUs

Dear Axel and Steve

Apologies for coming back to this only now.
Of course, the accuracy of the minimization should be adequate with respect to the desired result.
Nevertheless, I guess you might agree that obtaining different minimization results due to employing
a different number of MPI tasks can be quite startling - in particular for beginners.
I am going to dive deeper into parallel programing be the middle of next year and might just revisit
the propagation of these rounding differences in different parallel contexts as a challenging exercise.

Thank you very much again for your help