Stress artifacts in parallel mechanics solves

[ryar9534]

Describe the bug I am seeing some odd stress behavior in my poromechanical initialization of HI24L. In particular, there were some oscillations in stress on the boundaries between MPI ranks. On further testing I have narrowed it down to the mechanics solver, which seems to exhibit the same behavior. Here is an xml for a mechanics only sim on the underburden of HI24L: HI24L_mech_debug.zip

I ran with 1 rank (left) and 4 ranks (right), and slightly different results are obtained:

And the differences seem to line up with the MPI rank boundaries:

I have yet to see this effect on any other mesh, so perhaps its HI specific.