Debugging corrections and exact release

The following corrections are incorporated:

1. The unavailable original XVIII-B source is not represented as having been inspected or revised. The reconstructed benchmark is frozen explicitly.

2. The identical-particle factor $1/2!$, field factors of $\hslash$, source factor $g=\kappa /2$, and connected-cumulant factor $1/2$ are kept distinct.

3. “Once subtracted” refers to the bounded reduced densities $r_i$, not the growing densities ${\rho }_i=u^2{r}_i$.

4. The final tensor structures are polynomial; no rational projector is evaluated at $p^2=0$.

5. The contact classification is complete only for the declared transverse extension class. Nontransverse raw contacts are not silently admitted and then counted as physical parameters.

6. Zero-momentum normalization fixes degree-two and degree-four Taylor terms; the massive reference starts at degree six.

7. The anti-time-ordered kernel and its local contact have the sign needed for the CTP cancellation identity.

8. Vacuum spectral positivity is not promoted to instrument construction or observation.

9. The preparation residual is computed for an unrestricted transformation before any plateau restriction is applied.

10. The late detecting residual is separated from overlapping-support contact dependence.

11. The fixed mass is not treated as a massless completion, and preparation width is not a scheme parameter or a ultraviolet cutoff.

12. Linear transversality, a complete external-source inventory, and an inherited gravitational BRST/BV identity are not identified.