A fast algorithm for computing longest common subsequences of small alphabet size