Reduce number of all-to-all communications

At the moment we have multiple all-to-all communications.

Reduce the rebuild flag in engine_step(),
Reduce the minimal time end in engine_collect_timestep(),
Sum the number of updated particles.

Is there a way we could merge these somehow to avoid MPI barriers ?

The reduction of the number of updated particles is only for display to stdout. The reduction of the rebuild flag could be done at the same time as the reduction of the rebuild flag.

@pdraper would you be happy to look into this ? We'd like to avoid as much as possible the synchronization points within the code.

See !323 (merged).

To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information