Rewrote NodeWiseMeshPartitioner used in partmesh tool

The new algorithm is faster than the old one.

The example for time measurements was the partitioning of a cube mesh with 10.000.000 hex elements into 800 partitions. Included into the partitioning were two boundary meshes (left and right). The partmesh from master took circa 370 s, the partmesh with changes from this MR took 44 s.

1. Feature description was added to the changelog
2. Tests covering your feature were added? - All previously implemented tests are running
3. Any new feature or behaviour change was documented? - No new features, just an improvement of algorithm