Case optimization compiles out required MHD calculations in 1D/2D HLL cases

[StevenXue98]

Describe the bug
Case-optimized 1D and 2D MHD simulations using the HLL Riemann solver compile out required magnetic-pressure and total-energy-flux calculations. Several blocks in m_riemann_solver_hll.fpp are guarded by:

#:if not MFC_CASE_OPTIMIZATION or num_dims > 2

For a case-optimized 2D simulation, this condition is false. However, MFC uses three velocity and magnetic-field components for MHD even when num_dims is 1 or 2 (num_vels=3). Therefore, magnetic pressure remains zero and the MHD energy flux is not assigned, eventually producing NaNs.

To Reproduce

1. Check out the current master branch.
2. Run the standard 2D MHD rotor example with case optimization:

./mfc.sh run examples/2D_mhd_rotor/case.py \
  -t pre_process simulation \
  -n 1 -j 1 \
  --no-debug --no-gpu \
  --case-optimization

3. The simulation produces NaNs around time step 10:

NaN(s) in timestep output. 195 129 0 1 0 10 399 399 0

4. Run the same example without case optimization:

./mfc.sh run examples/2D_mhd_rotor/case.py \
  -t pre_process simulation \
  -n 1 -j 1 \
  --no-debug --no-gpu

5. The non-case-optimized simulation completes all 376 steps successfully.

Expected behavior
The case-optimized rotor simulation should complete successfully and produce the same numerical solution as the non-case-optimized build, within expected floating-point differences. Case optimization should change compilation and performance, not remove required MHD physics.

Screenshots
Not applicable.

Environment

• OS: Ubuntu 24.04.4 LTS, Linux 6.17.0-40-generic, x86-64
• Compiler: GNU Fortran 13.3.0
• MPI: Open MPI 4.1.6
• CPU build with --no-debug --no-gpu

Additional context
The affected HLL code in m_riemann_solver_hll.fpp includes:

• Relativistic MHD state calculations near lines 261 and 277.
• Non-relativistic magnetic pressure near line 285.
• Non-relativistic MHD energy flux near line 493.

Similar issues occur in m_riemann_solver_lf.fpp:

• Near line 254, case optimization removes the required relativistic-MHD state calculations in optimized 1D and 2D cases.
• Near line 395, case optimization removes the ideal-MHD energy-flux assignment in optimized 1D cases.

However, the LF Riemann solver is currently prohibited for MHD in case_validator.py, so these LF issues are not exposed through officially supported case configurations.