Refactoring of Shared Memory vs. OpenMP n-Body examples

* added Guideline Support Library as utility library
* refactored kernel code: loops, use of std::for_each and std::accumulate instead of hand-crafted code for better readability.
* unified coding style

See merge request !78

added Chapel n-Body example



See merge request !77

Added benchmarks (supermuc phase 1 and phase 2) for shared_mem_vs_open_mp examples



See merge request !76

use std::accumulate to accumulate the force instead of handcrafted loop. Use const auto& consequently.

reverted splitting the forall loops. This is not neccessary, because we calculate the forces based on an old copy

added second forall loop again, since updating location needs to be done in a separate step. To calc the force of one Particle, the locations of all other Particles (in the last simulationstep) are needed. Immediately updating the location leads to corrupt results

Repo: Manual reset to origin/master to clean up repo for merge request to upstream. Summary of changes: added input validation, fixed race conditions, major refactoring, improved readability and correctness, re-formatting of code.