Numerical Verification of a Curvature-Bound Bounce Cosmology: Effective-Field-Theory Control Pushes the Transition Below the Planck Scale

Version 1.5 — the preregistered verdict: no detection, g0 < 0.046. On 2026-09-03 the frozen v1.4 estimator was executed once on the Planck SMICA temperature map (file hashes preregistered; mask-filler seed fixed in advance; runtime 12.2 s). The axis-maximized statistic is 5.43 with global p = 0.27 against the measured 100,000-realization null distribution — no detection. The fixed-axis channel gives g0 = 0.019 ± 0.016, consistent with zero. Per the preregistered decision table, the result of this program is the upper limit g0 < 0.046 (95%), i.e. C ≲ 6.6×10-6 for the test-shear amplitude — the dedicated tilted-template limit the earlier indicative scale-independent number called for. A monopole/dipole-removed robustness variant, declared as post-registration, leaves every number essentially unchanged. The negative result is reported exactly as the preregistration required. CRBC is a phenomenological framework proposed by Ho Hyung Kim. Version 1.4 — the preregistered Planck-quadrupole analysis pipeline, frozen before data. This release closes every computational gate of the preregistration and timestamps the freeze: a torch/CUDA reimplementation of the pipeline's spherical-harmonic transforms (per-seed parity 1.1e-5 against healpy through the full estimator) runs each 100,000-realization null campaign in about six hours on one workstation. With the map's transfer function (5-arcmin effective beam and pixel window) forward-modelled in every simulation — matching it moved the estimator zero point by 1.1σ of the scatter — the frozen calibration is σ(g0)=0.0157, zero point 0.0436±0.0013, recovery slope 0.990±0.015, and a global 5% threshold of 10.09 for the axis-maximized statistic (2.63× the single-direction χ²1 value — the measured look-elsewhere penalty). Injections at an axis never used in training are recovered within one grid pixel (7.4°). The data-side procedure, including the mask-filler seed, is frozen in §18 of the preregistration document. No Planck map has been opened. CRBC is a phenomenological framework proposed by Ho Hyung Kim. Earlier (v1.3): Version 1.3 of the reproducibility preprint. This release adds derived-background spectrum re-integrations, an entropy-mechanism recheck, prior-work positioning, and a covariant-action audit. The results are explicitly conditional on the supplied phenomenological fluid background: the tested thermal Hagedorn, conformal-coupling, and background-reconstruction routes do not provide a covariant origin for the transition. A kinetic-dependent derivative coupling remains a candidate proxy only; a variational derivation, DHOST degeneracy, perturbative stability, strong-coupling control, and CMB likelihood analysis are not claimed.