Add LJ variant for HalfNeighbor list which uses data replication

[crtrott]

The OpenMP package in LAMMPS utilizes data replication to avoid write conflicts on the force array. This works very well for small amounts of threads. It may be worthwhile to explore to use a combination of data replication and atomics for higher thread counts (like on GPUs) to reduce the conflict rate while keeping the amount of data replication limited.

[stanmoore1]

There is a Kokkos ReductionView, see kokkos/kokkos#1225, that is being developed and is nearly finished. That can be used for this.

[stanmoore1]

See PR #21