Version 2.6 corrects a critical error identified in a post-publication audit of v2.5: the Fisher information calculation for LISA detectability (§7.4) used the pre-screening amplitude (Abruto=1.9148×10⁻⁸) instead of the post-screening amplitude (Afinal=1.46×10⁻¹⁶) that the rest of this work establishes as the physically observable value. Correcting for this raises the number of events required for 5σ detection from N₅=204 to N₅≈3.5×10¹⁸, making GRU qualitatively distinguishable (nGRU<1 signature, LISA amplitude window, unique quadrupolar anisotropy) but quantitatively indetectable by LISA in its primary mission with the screening mechanism as currently defined. This correction and its full derivation are documented transparently in new Appendix F.5. This version also corrects a non-isotropic sampling bug in the SO(3) symmetry-breaking ensemble test (conclusion unchanged: breaking ~10⁻⁸² at LIGO scale, confirmed analytically), and recomputes the Fermi-LAT GRB 090510 comparison ("C6") differentially across three methods and two energy bands, reporting the honest verdict of marginal, method-dependent tension rather than a single clean pass/fail. All other results — spectral dimension universality (ds(spine)≈1 across 2D, 3D real CDT, and toy 4D), the dimensional hierarchy (Δds≈4), the Root Invariance test (App. A.46), PPN/Solar System compatibility, and the CMB quadrupole screening mechanism — remain independently verified and unaffected by this correction. Full audit trail (scripts, JSON evidence, closing verification script) included in the package for independent reproducibility.
Alfredo Flores Cornejo (Thu,) studied this question.