In quantum field theory with a mass gap, why do states in the asymptotic future/past turn out to have a Fock space structure?

Asymptotically, particles and bound states in a QFT with mass gap behave as noninteracting particles, as the interaction decays exponentially because of the mass gap. Hence they are described by free fields. The physical S-matrix is between all these asymptotic states, and only these. Thus on the level of the asymptotic states we have particle democracy - elementary, composite and nonlocal (solitonic) particles appear on the same footing.

In particular, soliton states are asymptotically also describes by free fields, though LSZ is not directly applicable. (Textbooks conventionally just treat the case where there are neither bound states nor solitons. This includes Weinberg's QFT treatise; however, he at least acknowledges the problem in somewhat cryptic remarks on p.110 of his Vol. 1.)

If there is no mass gap (for example in QED), the situation is significantly more complicated, because then all asymptotic states describe so-called infraparticles, and asymptotic states are not Fock states but Fock states dressed with coherent states made from the massless particles. Working instead with Fock states (as in traditional QFT books) produces infrared divergences which must then be eliminated by ad hoc techniques. If coherent states are used, these infrared divergences do not appear.

On the other hand, unstable particles are visible asympotically only through their decay products, hence are not represented by Fock states. However, in approximate theories neglecting interactions that cause instability, one may regard the unstable particles as asymptotic particles.

Similarly, confined particles are represented asymptotically only through the bound states in which they appear, so they are also not represented by Fock states.