add method to set BCFunctor internal pointer

[cwentland0]

For assigning Schwarz boundary state and graph pointers, implemented by user BCFunctor class.

[fnrizzi]

can you please elaborate a bit more about what you mean by "boundary state and graph pointers"?
What is the purpose of doing this:

template <class T>
  void setBCPointer(::pressiodemoapps::impl::GhostRelativeLocation rloc, T* ptr) {
    m_bcFuncsHolder.setInternalPointer(rloc, ptr);
  }

I mean, what is ptr and what is it used for?

[cwentland0]

The Schwarz Dirichlet boundary conditions requires access to a boundary state, and a graph referencing the boundary cell index to the appropriate ghost cell index within the boundary state. This is the same as in the original Schwarz implementation (on the schwarz branch), which used

const auto bcIndex = m_graphBcs(gRow, 0);
      if (bcIndex == -1) {
        m_ghostLeft(gRow, 0) = m_state(uIndex+0);
        m_ghostLeft(gRow, 1) = m_state(uIndex+1);
        m_ghostLeft(gRow, 2) = m_state(uIndex+2);
        m_ghostLeft(gRow, 3) = m_state(uIndex+3);
      }
      else {
        m_ghostLeft(gRow, 0) = m_stateBcs(bcIndex+0);
        m_ghostLeft(gRow, 1) = m_stateBcs(bcIndex+1);
        m_ghostLeft(gRow, 2) = m_stateBcs(bcIndex+2);
        m_ghostLeft(gRow, 3) = m_stateBcs(bcIndex+3);
      }

in conjunction with routines that set those pointers

void setStateBc(state_type * stateBc){
    m_stateBc = stateBc;
  }

  void setGraphBc(graph_type * ghostGraph){
    m_ghostGraph = ghostGraph;
  }

The function you quoted is just a catch-all for whatever similar pointers the user-defined BCFunctor might need to set after initializing the problem.