This paper on closely packed orbit simulation seems to suggest that having the planets in first and second order resonances decreases the time till orbits cross each other. I'd got the impression that resonances were greater in stability the lower order they were, but if I'm reading this right, it seems like they get the most time in their simulation if they space each adjacent pair of planets at a certain multiple of their mutual hill spheres without regard to resonant ratios.
So if we had a system of planets in say 2:3 resonances through each pair, that would actually be more unstable (less time till orbit crossing) than if the planets were non-resonantly spaced? That seems backwards to how I understood this, though they do concentrate on the overlap of resonances, so perhaps its different for a whole chain of bodies rather than two. Then again, real systems like Trappist-1 exist which seem to be in long chains of near resonances or at least groups, so I'm not sure I understand this.