Determine the optimal depth of a sorting network on 18 channels: does a depth-10 network exist?

The depth of a comparator network is the number of parallel layers when comparators acting on disjoint channels are applied simultaneously; the optimal sorting-network depth $D(n)$ is the minimum depth over all sorting networks on $n$ channels. The values $D(n)$ are known exactly for $n \le 17$: in particular $D(16)=9$ and $D(17)=10$. For $n=18$ the optimal depth is not known - the best known sorting network on 18 channels has depth 11, while the only proven lower bound is $D(18) \ge D(17) = 10$. Determine $D(18)$: either exhibit a sorting network on 18 channels of depth 10 (giving $D(18)=10$), or prove that no depth-10 sorting network on 18 channels exists (which, with the known depth-11 network, gives $D(18)=11$).

Acceptance. FULLY RESOLVES: (a) a depth-10 sorting network on 18 channels - given as 10 layers, each a set of disjoint comparators - verified to sort all $2^{18}=262144$ binary inputs (settles $D(18)=10$); OR (b) a machine-checkable UNSAT certificate that no depth-10 sorting network on 18 channels exists (settles $D(18)=11$, since a depth-11 network is known). PARTIAL: a SAT/search advance that shrinks the space (e.g. exploiting the two-layer prefix normal form of Bundala-Zavodny) with reproducible artifacts. Provide the layered comparator list and a verifier, or the solver encoding and certificate.