look into making pbsdsh default bootstrap server for recent Intel MPI

[boegel]

advantages:

• doesn't require passwordless SSH to work (cfr. wrong SSH key permissions cause hang at MPI startup, no errors whatsoever #74)
• NUMA-aware placement of remote processes via Torque

example (with Intel MPI)

mpirun -bootstrap pbsdsh -np 40 ./mpi_helloworld

[boegel]

should actually use the pbsssh wrapper (see bin/pbsssh.sh)

[boegel]

relevant documentation:

https://www.vscentrum.be/cluster-doc/running-jobs/starting-programs-in-job#starting-a-single-core-program-on-each-assigned-core

http://docs.adaptivecomputing.com/torque/6-0-1/help.htm#topics/torque/commands/pbsdsh.htm

[Caylo]

the problem seems to be mostly the if/else under this:
https://github.com/hpcugent/vsc-mympirun/blob/develop/lib/vsc/mympirun/mpi/mpi.py#L820 where -rmk is prioritized over the launcher, which shouldn't happen.
(more info on that:
- http://www.failureasaservice.com/2014/04/mpich-and-mvapich-with-torque.html
- https://software.intel.com/en-us/forums/intel-clusters-and-hpc-technology/topic/278096)

HYDRA_LAUNCHER_NAME and HYDRA_LAUNCHER_EXEC are hardcoded into the schedulers, we should look into making this an option e.g. --launcher

also, the mympisanityscript needs some reviewing (see #80) so we can use it to debug/sanity check how processes are pinned

[boegel]

see #110

[boegel]

#110 is merged, but not perfect, see #114

[Caylo]

#114 fixed in #115