SCF relaxation hardcodes A/B points for z-axis rotation but does not validate `rot_axis`

Summary

do_hscf_solve() sets SCF anchor points as equatorial on x and polar on z (scf_r_A[0], scf_r_B[2]) and comments that this assumes z-axis rotation. There is no guard enforcing castro::rot_axis == 3, so users can run SCF with another axis and get inconsistent physics.

Location

• Source/scf/scf_relax.cpp:76
• Source/scf/scf_relax.cpp:77
• Source/scf/scf_relax.cpp:248
• Source/scf/scf_relax.cpp:382

Problem Details

Current setup:

scf_r_A[0] += castro::scf_equatorial_radius;
scf_r_B[2] += castro::scf_polar_radius;

Multiple later comments state the algorithm assumes rotation on the z-axis, but no runtime check enforces that assumption.

Impact

• SCF solution can be physically incorrect when rot_axis != 3.
• Silent misconfiguration risk: run completes with wrong equilibrium target.

Suggested Patch

Fail fast unless z-axis rotation is used for SCF:

diff --git a/Source/scf/scf_relax.cpp b/Source/scf/scf_relax.cpp
--- a/Source/scf/scf_relax.cpp
+++ b/Source/scf/scf_relax.cpp
@@   
     if (scf_polar_radius > scf_equatorial_radius) {
         amrex::Error("Polar radius must not be larger than equatorial radius for SCF relaxation.");
     }
+
+    if (castro::rot_axis != 3) {
+        amrex::Error("SCF relaxation currently assumes rot_axis = 3 (z-axis).");
+    }

Prepared by Codex

Start in Source/scf/scf_relax.cpp at lines 76-77, then inspect the related assumptions around lines 248 and 382. Add validation for the rotation axis before SCF relaxation proceeds, and run an SCF configuration with a non-z rotation axis to confirm it fails clearly while the z-axis case remains valid.