qbati/HAS-IDS: HAS-IDS v1.0.1

Maintenance and reproducibility release focused on improving experimental reproducibility and evaluation consistency. This release updates the corrected Distrinet-CIC-IDS2017 threshold-search implementation to evaluate observed calibration scores directly instead of a rounded score grid. This avoids rounding-induced degenerate all-Normal predictions during multi-seed evaluation while keeping the FPR constraint and tie-breaking logic unchanged. It also adds 10-run reproducibility scripts for UNSW-NB15 and corrected Distrinet-CIC-IDS2017: HAS-IDS/run_10seeds_unsw.py HAS-IDS/run_10seeds_dcic.py Documentation and citation metadata were refreshed for v1.0.3. No dataset split, model architecture, baseline implementation, or experimental protocol was changed.