Revision and debugging record

The manuscript begins with the unresolved physical input ledger. No diagnostic package is relabeled physical.

The adjoints, generic decision theorems, operator interfaces, and smooth benchmark have been moved out of the main calculation.

The singular-primitive theorem now explicitly assumes scalar scale regularity. Its proof uses absolute continuity, not merely an almost-everywhere derivative identity.

The main partition multiplies already differentiated seed amplitudes. Raw-sector motion terms are retained only in the separate raw-sector convention.

The $0R$ zeros for $47$, $5a$, and response rows are proved in precisely that post-adjoint convention.

Source derivatives remain ambient derivatives before incidence restriction. The $6$ derivative is never redistributed onto its incidence delta.

All three $47$ terms, including the derivative of $f$, and both $9$ root terms are retained.

The non-tip $5z$ chart coverage is not inferred from separation of $b-z$ alone. Its additional denominator condition is stated.

Cone-tip calculations use $T\mathrm{d}T\mathrm{d}\mathbf{m}$. Non-tip denominators are not extrapolated through $T=0$.

The calculated powers ${ϵ}^2$ and ${ϵ}^3$ describe internal localization. They are not assigned to the smoothing scale $\delta$.

The transition cross-check tests the translation construction against arbitrary compact profiles. It establishes the same geometric coefficients in both descriptions.

The two-parameter errors do not constitute a complete physical smoothing remainder. Complementary regions remain on the audit ledger.

The smooth-data theorem is a consistency benchmark, not a singular physical evaluation.

Finite-order estimates, physical authorization, reconstruction descent, wavefront control, and admission remain distinct.