Consolidate wave tests: fold correctness tests into convergence binaries (#1978)

### Description

This PR implements the proposal in #1916: each of the five wave
convergence binaries now supports a dual-mode `problem_main()`,
controlled by two TOML flags, `setup.run_convergence` (default `true`)
and `setup.run_sim` (default `false`). The four standalone correctness
binaries (FastWave, SlowWave, AlfvenWaveLinear, HydroWave) are deleted;
their CTets are re-registered against the corresponding convergence
binary with identical test names and labels, so CI coverage is
unchanged.

**Dual-mode design.** The two flags are independent: both can be `true`
simultaneously, in which case the exit status is OR'd. The default
(`run_convergence=true, run_sim=false`) means that bare invocations
(including the existing convergence CTets, which pass no `run_sim` flag)
are entirely backward compatible. The correctness TOMLs (FastWave.toml,
SlowWave.toml, AlfvenWaveLinear.toml, HydroWave.toml, HydroWaveFc.toml)
each get `setup.run_convergence=false` and `setup.run_sim=true`, turning
the correctness path on and the convergence path off.

**Correctness path.** The `run_sim` block reads wave geometry from the
TOML (`setup.num_modes_*`, `setup.angle_between_k_b0`), sets the
GPU-managed globals used by the IC kernel, constructs periodic BCs
explicitly, creates and evolves the simulation, then checks the error.
For the four MHD tests this is `sim.computeErrorNorm() <
setup.error_tol`; for HydroWave it is the fextract-based L1 norm that
the old standalone binary used. FastWave.toml gains the wave geometry
params that the old binary hardcoded (theta=90 deg, one x-mode);
HydroWave.toml gains `stop_time=1.0` and `max_timesteps=20000` that the
old binary hardcoded; HydroWaveFc.toml gains `cfl=0.1` because the old
binary hardcoded `sim.cflNumber_=0.1` and without it the one-step error
at the default CFL=0.3 marginally exceeds the 1e-8 tolerance.

**Convergence path.** The Richardson block is the old `problem_main()`
content, with one addition: `pp.query("nx_start", params.nx_initial)`
exposes the initial resolution as a TOML-overridable parameter alongside
the existing `setup.nx_max`.

I ran all binaries locally in both modes before opening this PR: all ten
test invocations (five binaries x two modes) pass.

### Related issues

- #1916: this PR implements the proposal described there.
- #1959: this PR was enabled by it; the N-dim Richardson extension was a
prerequisite for the dual-mode design to work correctly across 1D/2D/3D
setups.

### Checklist
- [x] I have added a description (see above).
- [x] I have added a link to any related issues (if applicable; see
above).
- [x] I have read the [Contributing
Guide](https://github.com/quokka-astro/quokka/blob/development/CONTRIBUTING.md).
- [ ] I have added tests for any new physics that this PR adds to the
code.
- [x] *(For quokka-astro org members)* I have manually triggered the GPU
tests with the magic comment `/azp run`.

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: AstroKriel

.github/workflows/mhd.yml

@@ -99,11 +99,12 @@ jobs:
       working-directory: ${{github.workspace}}/tests
       shell: bash
       run: |
-        ${{runner.workspace}}/build/src/problems/FastWave/FastWave \
+        ${{runner.workspace}}/build/src/problems/FastWaveConvergence/FastWaveConvergence \
           ../inputs/FastWave.toml \
           amr.max_grid_size_x=32 \
           amr.blocking_factor_x=16 \
           mhd.emf_compute_scheme=FelkerStone2017 \
+          mhd.emf_averaging_scheme=LondrilloDelZanna2004 \
           amrex.init_snan=1 \
           amrex.fpe_trap_invalid=1 \
           amrex.fpe_trap_overflow=1 \
@@ -113,7 +114,7 @@ jobs:
     - name: Test
       working-directory: ${{runner.workspace}}/build
       shell: bash
-      # Execute the MHD tests while skipping the slow HydroWave case
+      # Execute the MHD tests while skipping the slow HydroWaveConvergence case
       run: canary run --output-on-failure --workers=2 -k 'MHD and not HydroWaveConvergence' .
 
     - name: ccache stats

inputs/AlfvenWaveLinear.toml

@@ -34,6 +34,9 @@ hydro.rk_integrator_order = 2
 hydro.reconstruction_order = 5
 hydro.use_dual_energy = 0
 
+setup.run_convergence = false
+setup.run_sim = true
+setup.error_tol = 0.005
 setup.num_modes_x = 1
 setup.num_modes_y = 0
 setup.num_modes_z = 0

inputs/EntropyWave.toml

@@ -0,0 +1,42 @@
+# *****************************************************************
+# Problem size and geometry
+# *****************************************************************
+geometry.prob_lo = [0.0, 0.0, 0.0]
+geometry.prob_hi = [1.0, 1.0, 1.0]
+geometry.is_periodic = [1, 1, 1]
+
+# *****************************************************************
+# VERBOSITY
+# *****************************************************************
+amr.v = 0  # verbosity in Amr
+
+# *****************************************************************
+# Resolution and refinement
+# *****************************************************************
+amr.n_cell = [128, 8, 8]
+amr.max_level = 0  # number of levels = max_level + 1
+amr.blocking_factor = 8  # grid size must be divisible by this
+amr.max_grid_size = 128  # maximum grid size to create 4 FABs of 25 cells each
+
+do_reflux = 0
+do_subcycle = 0
+do_tracers = 1
+plotfile_interval = -1
+
+cfl = 0.3
+stop_time = 1.0
+max_timesteps = 10000
+
+hydro.rk_integrator_order = 2
+hydro.reconstruction_order = 3
+hydro.use_dual_energy = 0
+
+mhd.reconstruction_order = 3
+
+setup.run_convergence = false
+setup.run_sim = true
+setup.error_tol = 0.002
+setup.num_modes_x = 1
+setup.num_modes_y = 0
+setup.num_modes_z = 0
+setup.angle_between_k_b0 = 0.0

inputs/EntropyWaveConvergence.toml

@@ -25,7 +25,6 @@ plotfile_interval = -1
 
 cfl = 0.3
 
-
 hydro.rk_integrator_order = 2
 hydro.reconstruction_order = 3
 hydro.use_dual_energy = 0

inputs/FastWave.toml

@@ -36,5 +36,10 @@ hydro.rk_integrator_order = 2
 hydro.reconstruction_order = 5
 hydro.use_dual_energy = 0
 
-mhd.emf_compute_scheme = "Balsara2025"
-mhd.emf_averaging_scheme = "LondrilloDelZanna2004"
+setup.run_convergence = false
+setup.run_sim = true
+setup.error_tol = 0.002
+setup.num_modes_x = 1
+setup.num_modes_y = 0
+setup.num_modes_z = 0
+setup.angle_between_k_b0 = 90.0

inputs/HydroWave.toml

@@ -22,3 +22,8 @@ do_reflux = 0
 do_subcycle = 0
 do_tracers = 1
 plotfile_interval = -1
+stop_time = 1.0
+max_timesteps = 20000
+
+setup.run_convergence = false
+setup.run_sim = true

inputs/HydroWaveFc.toml

@@ -23,9 +23,14 @@ amr.blocking_factor_z = 8
 
 do_reflux = 0
 do_subcycle = 0
+cfl = 0.1
 
 plotfile_interval = 1
 checkpoint_interval = -1
 stop_time = 1.0
 max_timesteps = 1
 do_tracers = 1
+
+setup.run_convergence = false
+setup.run_sim = true
+setup.error_tol = 1.0e-8

inputs/SlowWave.toml

@@ -41,6 +41,9 @@ mhd.reconstruction_order = 5
 mhd.emf_compute_scheme = "Balsara2025"
 mhd.emf_averaging_scheme = "LondrilloDelZanna2004"
 
+setup.run_convergence = false
+setup.run_sim = true
+setup.error_tol = 0.002
 setup.num_modes_x = 1
 setup.num_modes_y = 0
 setup.num_modes_z = 0

regression/quokka-tests.ini

@@ -191,7 +191,7 @@ ignore_return_code = 0
 
 [LinearAlfvenWave-GPU]
 buildDir = .
-target = AlfvenWaveLinear
+target = AlfvenWaveLinearConvergence
 inputFile = inputs/alfven_wave_linear_regression.toml
 dim = 3
 cmakeSetupOpts = -DAMReX_GPU_BACKEND=CUDA -DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache -DCMAKE_CUDA_COMPILER_LAUNCHER=ccache