perhaps one should mention in addition,
that it is a valid consideration to make
nested loops over the same neighborlist,
since 3-body terms in most of the potentials
in the manybody package are all shortranged
interactions and thus it is a valid assumption
that the third atom is as likely within the
neighbors of the first atom as the second.
dihedral (i.e. 4-body) terms are a bit more
tricky in that respect.