Calibration-Family Limits on Directional Reconstruction from Public ST7-DRS Closed-Loop Telemetry

This preprint presents a model-conditional analysis of calibration-family limits on directional reconstruction from public ST7-DRS closed-loop telemetry. The journal-facing analysis treats the published HF/LF calibration discrepancy as a conditional calibration-family sensitivity rather than as a branch-selection problem. The central HF/LF directional separation is approximately 3.20°, while branch-conditional 99% uncertainty regions remain separated by more than 3.089° across the tested block lengths. The accompanying reproducibility release includes the frozen analysis code, exact non-FITS inputs, replay outputs, cryptographic manifests, and provenance required to reproduce the analysis. A complete end-to-end replay of the frozen PHYS-P9 audit was performed from all six declared inputs, with all inputs matching their archived SHA-256 hashes and byte sizes. The two large public NASA/HEASARC FITS products are referenced by exact filenames, archive locations, byte sizes, and SHA-256 hashes rather than duplicated in the compact archive.