Third-order upwind FD scheme

[valeriabarra]

Since a 1st-order upwind FD scheme is too dissipative (edit: typo), we want to add a 3rd-order upwind FD scheme.

This can then be used with limiters in hybrid domains (both 2D hybrid domain, 1D CG x 1D FD, and 3D hybrid domain, 2D CG x 1D FD).

Here are a couple of references:

• Book: Hundsdorfer and Verwer - 2003 - Numerical Solution of Time-Dependent Advection-Diffusion-Reaction Equations (equations (3.27) & (3.28) for regular unstaggered upwind third-order finite differences stencil)
• Paper on staggered reconstruction: Pressel et al. - 2015 - Large-eddy simulation in an anelastic framework with closed water and entropy balances doi: 10.1002/2015MS000496 (equations (60)-(63))

[valeriabarra]

@simonbyrne who can look into this?

[valeriabarra]

This was merged, but it currently works only with periodic meshes. If anyone wants to work on adding the boundary support, they can feel free to re-open this issue or create a new one.