Full-Space Riesz Margin Diagnostics for Anisotropic Gaussian Channels

We develop a full-space Gram--Riesz margin diagnostic framework for anisotropic Gaussian channels in the probability space L² (R³, dμ0). The harmonic anisotropy multiplier is treated as an unbounded positive quadratic form on an explicit finite-cell form domain, replacing the bounded-window pointwise-envelope argument by a full-space form-compression viewpoint. The numerical layer uses a Cartesian weighted-Hermite total-degree execution core and a manifest-linked replay pipeline. A finite-depth diagnostic at beta=1, epsilon=0. 60, and L=6 yields lambdaₘin = 0. 0290032510108 after quadrature refinement to Q=256, with witness mass 0. 991055491 in the low-multiplier region aₑpsilon <= 0. 25. This result is reported strictly as a replay-stable, escape-associated finite-depth loss of lower Riesz margin; it is not an infinite-rank Riesz-collapse theorem, not an asymptotic scaling law, and not a cutoff-free Coulomb BBGKY closure claim.