Papers 75, 76, and 77 identified three ways a system can be structurally unable to converge, even under favorable conditions. This paper answers the question those three papers left open: are those the only ways? The answer is yes — and this paper proves it. P78 establishes the formal closure of the CM (Collapse Mode / Failure Mode) namespace under Zero Leap Theory. The proof has three components. CM Completeness (P78.T1): the ZLC parameter space {kappa, K, Pi} is three-dimensional binary, producing exactly eight cells. One is ZLC-positive. Three are pure ZLC-negative, covered by CM-001 (Idolatry Failure, P75), CM-002 (Pharisee Failure, P76), and CM-003 (Atomization Failure, P77). No independent fourth structural parameter can produce ZLC-negativity while all three ZLC conditions hold — this follows directly from ZLC sufficiency (P72.T1): if kappa ≥ kappa_min, K > 0, and Pi > 0 all hold, R/I → 0 is guaranteed. CM Orthogonality (P78.T2): the three failure dimensions are structurally independent. None implies any other under the axiom set. Compound states are empirically common but structurally contingent, not logically necessary. Compound Decomposition (P78.T3): every compound ZLC-negative state is uniquely decomposable into a superposition of active pure modes, with a canonically ordered intervention sequence — kappa → K → Pi — derived from admissibility dependencies. When restoration is structurally feasible, that order is provably optimal. When kappa < kappa_irr (P77.T3), the compound state may be irreversible under NOT-P3; the theorem is explicitly conditional. The paper also formally registers the P50 versioning policy for the CM namespace and closes the ZLT-CM registry: ZLT-CM-004 will not be defined. No new axioms, gates, or consequence modes are introduced. This is Paper 78 of the Zero Leap Theory series and the formal closure of the Structural Failure Mode Triad (P75–P76–P77–P78).