Verify dataset against remote checksum after download

[bencehezso]

Summary

• Verify the downloaded dataset against the remote SHA-256 checksum before extracting; discard the file on mismatch so a corrupted or truncated download can never populate data.db.
• Replace the weak "dataset folder is non-empty" guard with an explicit check that datasets/data.db exists and is non-empty.
• Use sys.exit(1) instead of the builtin exit(1) in utils/data.py.