First pass at core GR infrastructure

[JonathanGorard]

This PR adds prototype support for general relativistic/curved spacetime hydrodynamics to the moment app (in the form of the wv_gr_euler equation object), along with core infrastructural support for broader numerical relativity functionality in Gkeyll. It introduces the gr_spacetime object, with two specific instances (gr_minkowski, for special relativity, and gr_blackhole, for both static and spinning black holes in Kerr-Schild coordinates) implemented for the time being. It also adds basic unit tests for flat and curved spacetime functionality, and regression tests for relativistic hydrodynamics in both the SR and GR limits (including full Bondi-Hoyle-Lyttleton accretion tests onto both static and spinning black holes). Although all current test cases use stationary spacetime metrics (either Minkowski or Kerr-Schild), the underlying infrastructure has been designed to support fully dynamic spacetimes, as evolved through the Bona-Massó equations or their equivalent - to this end, the wv_gr_equation object actually consists of 29 equations, with the usual 5 for the fluid variables, 18 for the (spatial) metric and inverse (spatial) metric components, 1 for the (spatial) metric determinant, 1 for the ADM lapse function, 3 for the ADM shift vector components, and 1 for the excision boundary condition. For now, only the first 5 are actually evolved. Currently, the gauge conditions on the lapse and shift parameters are enforced as part of the gr_spacetime object, but going forward (and especially as we transition to dynamic spacetimes) we will likely want to solve the constraint equations via the moment app directly, and so we have designed the wv_gr_euler object with this in mind.

This functionality should still be considered highly experimental, and, in particular, the fluid solver currently supports only Lax fluxes (with a partially-implemented general relativistic Roe solver that currently does not work). No one should (yet) be basing real research results on this code! Nevertheless, I would like to merge this PR sooner rather than later, for several reasons:

• This new functionality does not/should not conflict with any aspect of the Gkeyll codebase, and so is perfectly safe to merge.
• We (@JunoRavin and I) would ideally like to begin prototyping a new design for the handling of sources in the moment app, and having the GR hydrodynamics code already merged into main will make it much easier to do this.
• I would also like to begin gradually deprecating/removing the existing SR hydrodynamics code in the moment app, migrating those features and tests over to the new GR hydrodynamics system in the process. Of course this will not happen until the Roe solver is fully robustified and validated.
• The utility functions in the gr_spacetime object for calculating Christoffel symbols, extrinsic curvature tensors, etc. may prove useful in other parts of the Gkeyll code (e.g. for the work that @johnson452 is doing on curved spacetime kinetics), and these have been accordingly validated via reasonably extensive unit tests for both flat and black hole spacetimes.
• I would gradually like to repurpose the gr-prototyping branch for forthcoming work on dynamic spacetime functionality and other, related things.