Saved clock-boundary review

Both failed fits place all 119 receiver queries close to internal integration boundaries. The reconstructed clock arrays match their original saved hashes. This identifies a useful diagnostic target; it does not establish a derivative error or the cause of the failure.

The offset is near −2/300 seconds, an internal clock boundary rather than the permitted offset limit. The recorded fresh correction fails cost stability while passing the stationarity and position-change checks. A later qualification should test both timing branches and known optima before changing any solver.

Twenty-two software checks pass. This review reads two saved fits and generates only clock metadata: zero new noise draws, optimizer starts or native calls. Validation remains blocked; the original 53 aircraft comparisons are unchanged.

Per-query CSV · Fixed review scope · Calibration outcome · Next analyses.