within the Journe watch, despite what you say about drag effects between the two balances. It has been described elsewhere on ThePuristS (from memory, by Felipe) how a Journe watch's two movements will start to beat in synchrony a few minutes after winding; and remain that way (as evidenced by continued synchrony of the seconds hands). Presumably there are acoustic waves being generated by the balances which are either not affected, or little affected, by interference patterns at the closest point.
Acoustic resonance, Thomas, is no different from what you term "mechanical" resonance. The coupling medium is merely a gas instead of a solid. In fact, from my understanding of the Haldiman movement, I doubt whether it can really be termed a resonant device, as the two mainsprings appear to be directly coupled. I can't imagine, for instance, how one side of the device could oscillate while the other was still. This seems to me to be no more a matter of resonance than the interaction between, say, the second and third wheel of a normal movement.
While I take Thomas' point (from other threads, esp. the Haldiman thread) about the dubiousness of claims for improved timekeeping in supercomplications, based on lack of demonstrable benefit in practice, it is nevertheless worth discussing the theoretical benefits, as such discussion can at least lead to the formulation of an experimental design to test any hypotheses which arise. (It's then up to the manufactures themselves to decide whether such tests are worth funding.)
In the case of a resonance watch, I submit that there could theoretically be a benefit in accuracy derived from the averaging effects produced by coupling of the two movements.