GPU volume fraction refinement

[rhornung67]

Develop capabilities to support GPU execution of material volume fraction initialization on refined meshes in AMR context.

• [x] Initial discussion with application team(s) about capability needs
• [ ] Refine and agree on requirements with app team
• [ ] Develop API and vet with app team
• [ ] Develop initial implementation for CPU execution and compare to existing app capability (accuracy, performance). Based on test problems from app team.
• [ ] Develop GPU capability and compare solution with runs on CPU
• [ ] Develop Axom tests as needed for capabilities

[BradWhitlock]

I'm just making a note here that it looks like the target host code builds Axom without MFEM, and with a serial RAJA. This impacts our ability to take MIR topological output and compute volume fractions using the intersection shaper. The intersection shaper currently uses MFEM. On a GPU-enabled platform, this may be different.