The semi-empirical mass formula's pairing term is attractive only for even-even nuclei, see for example, Miriam Hein et al. $2022$ Eur. J. Phys. $43$ $035801$. That is, it is pairing between protons or pairing between neutrons. This is the isospin $T=1$ channel, where the $S=0$ interaction is most attractive. Your statement that $S=1$ is most attractive is true for $T=0$ i.e. the deuteron channel, but this is not the pairing contained in the semi-empirical mass formula term.