GitLab

* PME_ORDER threads process one atom;
* PME_ORDER threads access consecutive addresses;
* No need to permute z indices with zindexTable;
* finishSpreadCharge is needed only with fixed point charge spreading;