The Coleman-Mandula theorem (CMT) does not rule out theories with massless particles. What it rules out is a theory with only massless particles. If you only have massless particles you either:
- End up with a free theory. This theory has trivial S-matrix, thus, the CMT does not apply here
- Have conformal symmetry. If you have conformal symmetry you cannot strictly speak about particles (there is no localized state) and there is no S-matrix, since there is no asymptotic free states.
This is the reason why they add a mass gap. But you can have theories with massless particles and massive particles.
The second question can be answered as follows: Suppose you have a symmetry that is not a direct product of Poincaré and an internal symmetry. This means that you can apply to a particle state of mass $M_1$ at position $x_1$ your symmetry and transform it in a state of mass $M_2$ (or just a different particle) at position $x_2$. This looks like a long range force. The CMT assumes local forces, otherwise you cannot assume asymptotic free states.