ci: retrigger after Batch 2 timeout flake

镜像里没有 sqlite3 CLI，原来的 .backup 命令 15 天前就静默失败了。
daily-minio-backup 用 cp 就够，但 hourly 每小时跑要求更强一致性，
改用 Python 内置 sqlite3 的 .backup() API。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

deploy/backup-cronjob.yaml

@@ -18,12 +18,14 @@ spec:
                 - sh
                 - -c
                 - |
+                  set -e
                   BACKUP_DIR=/data/backups
                   mkdir -p $BACKUP_DIR
                   DATE=$(date +%Y%m%d_%H%M%S)
-                  # Backup SQLite database using .backup for consistency
-                  sqlite3 /data/oil_calculator.db ".backup '$BACKUP_DIR/oil_calculator_${DATE}.db'"
-                  echo "Backup done: $BACKUP_DIR/oil_calculator_${DATE}.db ($(du -h $BACKUP_DIR/oil_calculator_${DATE}.db | cut -f1))"
+                  DST="$BACKUP_DIR/oil_calculator_${DATE}.db"
+                  # Consistent snapshot via Python's sqlite3 .backup API (sqlite3 CLI not in image)
+                  python3 -c "import sqlite3; s=sqlite3.connect('/data/oil_calculator.db'); d=sqlite3.connect('$DST'); s.backup(d); d.close(); s.close()"
+                  echo "Backup done: $DST ($(du -h $DST | cut -f1))"
                   # Keep last 48 backups (2 days of hourly)
                   ls -t $BACKUP_DIR/oil_calculator_*.db | tail -n +49 | xargs rm -f 2>/dev/null
                   echo "Backups retained: $(ls $BACKUP_DIR/oil_calculator_*.db | wc -l)"