now always throws error if a bond type changes, but both of of the involved atoms are too close to edge of template

This is for consistency with other parts of LAMMPS where
such variables have "flag" in their name. Also reduces
confusion with have_gpu_direct() function.

When we can safely detect, that GPU-direct is not available,
change the default setting of Kokkos::gpu_direct_flag from 1 to 0

this adds a local wrapper function have_gpu_direct(), that informs about
the support for GPU-direct, if possible and add a warning message in case
it cannot be detected or is not available and using more than 1 MPI rank.

- improvements documentation (dmi and exchange)
- correction error cross product in pair_spin_dmi.cpp
- implementation mech. part in pair_spin_dmi.cpp
- correction in all pairs: init_one for [j][i] couples
- correction in atom_vec_spin.cpp: index error in read_data
- some improvements in pair_spin_dmi.cpp and pair_spin_magelec.cpp

after an exception "all bets are off",
i.e. the user should be allowed to do anything to recover.
through setting Update::whichflag to 0, the guard against running commands during a run is removed.

- document previously undocumented OpenCL tune settings
- implement OpenCL platform selection through prefixing the device type with the platform id separated by a colon
- allow passing custom tune parameters though postfixing the device type with the 13 tuneable parameters separated by commas
- remove an extra clear() that would delete device properties structs an cause LAMMPS to output garbage strings

The fix_shardlow_kokkos.cpp code had already factored out a redundant
sqrt() calculation in the innermost loop of ssa_update_dpde().  This
changeset propagates an equivilent optimization to:
  fix_shardlow.cpp
  pair_dpd_fdt_energy.cpp
  pair_dpd_fdt_energy_kokkos.cpp
The alpha_ij variable was really just an [itype][jtype] lookup parameter,
replacing a sqrt() and two multiplies per interacting particle pair
by a cached memory read.  Even if there isn't much time savings, the
code is now consistent across the various versions.