`riemann_solvers.H`: multigroup radiation state storage overwrites `QRAD` instead of writing `QRAD+g`

Location

Source/hydro/riemann_solvers.H:174-176

Problem

In compute_flux_q(..., store_full_state=true) under #ifdef RADIATION, the loop over groups writes:

for (int g = 0; g < NGROUPS; g++) {
    qgdnv(i,j,k,QRAD) = qint.er[g];
}

This overwrites the same component (QRAD) every iteration, leaving only the last group value and not populating QRAD+1... components.

Impact

• Incorrect full-state Godunov interface data for multigroup radiation.
• Derived quantities / diagnostics / downstream logic reading QRAD+g can see corrupted or stale values.

Steps to Reproduce

1. Run with RADIATION enabled and NGROUPS > 1.
2. Trigger path where store_full_state is true.
3. Inspect qgdnv(i,j,k,QRAD+g) components.
4. Observe only QRAD is updated repeatedly; higher groups are not written correctly.

Expected Behavior

Each radiation group should be stored in its own component.

Actual Behavior

All groups are written to QRAD only.

Proposed Fix

for (int g = 0; g < NGROUPS; g++) {
    qgdnv(i,j,k,QRAD+g) = qint.er[g];
}

Prepared by Codex

Start at Source/hydro/riemann_solvers.H:174-176 in compute_flux_q(..., store_full_state=true) under RADIATION, and inspect how the loop over groups populates qgdnv. Confirm that each group is stored in its corresponding QRAD+g component, then validate with RADIATION enabled and NGROUPS greater than one by inspecting the full-state interface data.