compute stress/atom of specific pair

How hard would it be to modify compute group/group to include stress?

could be done, at least for the true pairwise potentials
that compute group/group supports

Steve