AIJ Niigata (Case E) wind-tunnel benchmark: code_saturne vs OpenFOAM

URANS (k-ω SST, all-y+ wall treatment) cross-comparison of code_saturne v9.1 and OpenFOAM v2512 (pimpleFoam) on the same 1,368,424-cell snappyHexMesh mesh (no prism layers - isotropic near-ground refinement instead), against the AIJ UWE Benchmark Case E (Niigata) wind-tunnel experiment: a real urban district (a 60 m tower and two 18 m buildings), 1:250 scale, wind sector W, Re ≈ 17,000 (model scale). Validated against 78 of the 80 published pedestrian-height (z=2 m full scale) velocity-ratio measurements: R=0.783/RMSE=0.196/bias=-0.049 (code_saturne) vs R=0.774/RMSE=0.204/bias=-0.052 (OpenFOAM). Includes case definitions for both solvers (steady RANS restart into URANS for OpenFOAM; setup.xml and inlet-profile user function for code_saturne), the shared mesh (MED and native OpenFOAM formats, plus the snappyHexMesh generation recipe and logs), the probe/field results actually used for the published comparison, the AIJ Case E reference dataset (Tominaga, Kikumoto & Okaze, CC-BY 4.0), and the post-processing scripts. Corrects a field-source mismatch present in the originally published field-map figure (see README). See README.md for the full modelling table, results table and reproduction instructions.