Non-optimized post-processing may be broken

[mfarthin]

@kajiglet (internal erosion) and @tweigand (density dependent flow) have seen some velocity fields that should have been 1d but had some corner effects near the boundary.

We need to check Larson-Nicklasson examples to make sure still working at least

[mfarthin]

The single-phase flow example in proteus-sandbox/LN_Examples/Ex1-Obstacles_D1 "look ok" although I don't have an actual regression test. @tweigand, are you still seeing problems with the post-processed velocities in your brine example?

[tweigand]

The velocities are fine now for the brine example.

[cekees]

I found that pwl-bdm does't work in parallel in general, but I'm not sure it was ever designed to. pwl-bdm-opt does. I think we just need to update/cleanup/test proteus.PostProcessing. I'd like to add ('grad(velocity)',ci) and ('div(velocity)',ci) calculations as well, when we do that.

[adimako]

@cekees, perhaps close this one, and re-open if it becomes an issue again