On the equivalence between some projected and modulus-based splitting methods for linear complementarity problems