Creating a clean benchmarking system

[fluidnumerics-joe]

• Pixi environment management
• Pre-commit hooks
• All benchmarks need to be put under benchmark/ subdirectory
• We need to be able to ingest input datasets from public urls (we can use pooch to easily do this)
• parameterize choices with sensible defaults ( number of particles, time step size, simulation duration, advection scheme, ds.load or not, )