First width audit preserved
Two of six checks passed; four rejected the parent's valid outward-rounded width reports. The new checker had expected ordinary floating-point subtraction. All six charged identities and their original source freeze are preserved. No prediction was changed and no trajectory integration or aircraft fit ran.
The parent reports the exact endpoint difference rounded upward. A separately frozen corrected audit retains that convention and records its small reporting increment explicitly. The first study is complete and cannot be resumed.