Compare average code length with source entropy to compute redundancy (avgLen - H) / H; lower redundancy means a better code.
冗餘越低編碼越優,理想情況趨近 0。