Keeping it sparse: Computing Persistent Homology revised