On the need of precision in the calculation of the LCAO density matrix of polymers