Version 2.8.1 (2026-06) — pre-registration lock-in patch. The Layer-C forward-test pre-registration lock-in DOI, previously marked 'forthcoming', is now instantiated: the immutable pre-registration (vulnerability-score formula, test statistic, null distribution, decision rule) is anchored to this deposit's Zenodo concept DOI 10.5281/zenodo.20095776 and the Research Lab Git-mirror commit. No results or numbers change. (Residual internal version/process annotations are being removed progressively in subsequent revisions.)Version 2 (paper v2.8, 2026-06) — major revision of the v1 deposit (v1 remains citable at its version DOI). (1) The v1 thesis that an absent public backstop implies T_retail approximately 0 with catastrophic amplification is revised: the realized Q1-2026 non-traded-BDC redemption episode was absorbed in roughly six weeks via private substitutes (sponsor capital, prospectus gating discretion, legal pro-ration), with no fund failure. (2) The time-series contagion statistics (DCC-GARCH persistence 0.967, Forbes-Rigobon adjusted correlation 0.43, Diebold-Yilmaz connectedness 68%) are reclassified as v1 post-hoc characterizations — correcting a v1 conflation of the DCC persistence with a correlation level — while the paper's own v2 inference (Spearman with a 10,000-permutation null, p=0.110; bootstrap CI; simulated power MDE80=0.65; cross-sponsor joint-saturation Monte Carlo P(≥2) approximately 86%) is fully reproduced by deterministic, seeded code. Honest epistemic-status framing throughout.Extends the K-A-T framework to retail distribution channels. Three formal propositions: (i) retail funding constraints amplify procyclically through behavioral contagion and information asymmetry (K_retail); (ii) redemption cascades in semi-liquid vehicles create amplification mechanisms structurally analogous to bank runs but without deposit insurance (A_retail); (iii) absent lender-of-last-resort implies near-zero policy backstop (T_retail ~ 0). Empirical analysis on novel SEC EDGAR XBRL panel of 10 publicly-traded BDCs (138 quarterly observations, 2020-2025), plus event-snapshot data for non-traded vehicles. The March 2026 gate wave (BCRED 7.9%, HLEND 9.3%, Cliffwater 14%) yields Forbes-Rigobon adjusted correlation 0.43 (rejecting no-contagion at 1%), DCC-GARCH persistence beta = 0.967 (half-life 36 trading days), and Diebold-Yilmaz connectedness 68.0% (ARCC dominant net transmitter +12.6pp). Placebo: semi-liquid BDC connectedness 75.6% exceeds daily-liquid ETF 64.5%. Proposes Retail Private Credit Gateway (RPCG) regulatory framework: mandatory liquidity buffers (15-20% NAV), standardized valuation disclosure, ERISA-compliant suitability, FSOC designation authority for systemically important semi-liquid vehicles.