Function build_equivalence_queries_gpu 

pub fn build_equivalence_queries_gpu(
    phi: &GpuCnf,
    circuit: &GpuXgcf,
    provider: &Arc<CudaKernelProvider>,
) -> Result<GpuEquivalenceQueries>

Build the two equivalence-check queries on GPU:

• q1 = φ ∧ ¬C
• q2 = C ∧ ¬φ

This helper exists so tests and tooling can inspect query CNFs without duplicating kernel orchestration logic.