Add constant derivative boundary condition and expose GenericBC

[lroberts36]

PR Summary

This PR adds:

• The option of a constant derivative boundary condition (which I needed when experimenting with evolving the vector potential).
• The option of a constant value boundary condition (which I also needed at some point)
• Slightly refactors the internal pieces of GenericBC to use sparse packs and allow for the selection of which variables to apply boundary conditions on
• Moves GenericBC to a header so that downstream codes can compose the generic boundary conditions for subsets of variables. This should make it easier to create user defined boundary conditions that behave differently for different variables, e.g.

 pman.app_input->boundary_conditions[parthenon::BoundaryFace::inner_x1] = 
    [](std::shared_ptr<MeshBlockData<Real>> &rc, bool coarse) { 
      using namespace parthenon; 
      using namespace parthenon::BoundaryFunction; 
      GenericBC<X1DIR, BCSide::Inner, BCType::Outflow, variable_names::any>(rc, coarse); 
      GenericBC<X1DIR, BCSide::Inner, BCType::ConstantDeriv, myvar_t>(rc, coarse); 
      GenericBC<X1DIR, BCSide::Inner, BCType::Fixed, myvar2_t>(rc, coarse, 1.0); 
    };

• Additionally, it fixes a small bug in sparse packs, ensures that the vector components of fields in a sparse pack are correctly set, exposes MeshData level ApplyBoundaryConditions tasks, and removes some unused code.

PR Checklist

• Code passes cpplint
• New features are documented.
• Adds a test for any bugs fixed. Adds tests for new features.
• Code is formatted
• Changes are summarized in CHANGELOG.md
• CI has been triggered on Darwin for performance regression tests.
• Docs build
• (@lanl.gov employees) Update copyright on changed files

[Yurlungur]

This is a nice improvement. Thanks for adding it. Assuming everything works, I'm in favor. 👍

[lroberts36]

This at least partially closes #376

[lroberts36]

I used this in a downstream code that does CT using a vector potential (which requires a constant derivative boundary condition to give outflow boundary conditions for B, while the other conserved variables have an outflow boundary condtion) and everything seems to work well.

[brryan]

LGTM, this is nice to have. I think the if constexpr change would be worthwhile performance-wise if you agree and don't mind doing that before merge.

I thought about whether we should have a bunch more functions like ConstantInnerX1 etc. but I'm not sure that's a good idea. As in your use case different variables often have different BCs. Maybe the right way to do this would be to tag variables with their boundary conditions in their metadata, and use that to construct a BC function at runtime? Not a task for this PR of course.

[Yurlungur]

As in your use case different variables often have different BCs. Maybe the right way to do this would be to tag variables with their boundary conditions in their metadata, and use that to construct a BC function at runtime? Not a task for this PR of course.

I have a plan for this... metadata is one option, per-package BCs would be another. I've been meaning to refactor the code to add this forever, but I keep putting it off...

[pdmullen]

LGTM 💯