Splitting one job into smaller pieces leads to loss of some trajectories

[Alexsp32]

Splitting a job with 100 trajectories loses some trajectories from the total after splitting when total_trajectories % processes !=0.
Probably something I messed up in the routine to handle these cases.