Opportunities to improve consistency in sheath boundary #459
Description
We had an effort to improve sheath BC consistency here: #248
There are two more improvements that have been identified.
The first is based on Tom Body's thesis on the same for GRILLIX:
-Conserve Ni*Vi into sheath
-Conserve Ni*Ti into sheath
-Conserve Ne/Ni into sheath (not sure how else to deal with quasineutrality).
The reason we don't conserve Ni*Vi is because we extrapolate Ni and set Vi to be equal or greater than Bohm. Making this more consistent reduced boundary zigzags in GRILLIX.
The second improvement was identified by @dschwoerer. We allow exponential extrapolation of quantities into the sheath boundary to calculate the guard cell value, but then to calculate the value at the sheath boundary, we do regular linear interpolation. This seems inconsistent.
I started working on this in #257.