[staggered][fluxVars] Adapt treatment of interior boundaries

* so far, interior boundaries (e.g., flow around a block)
  were always interpreted as Dirichlet (no-slip) boundaries
* the reduced distance used in the calculation of the gradient
  was already correctly incorporated, we just document it better here
* this concerns the gradient for the shear stress term, e.g., du/dy
  at an edge within the domain