37 / C5-A: Record-Window Renormalization, Compensation-Motif State Space, and Metadata Compactness

C4 has already compressed blow-up's asynchronous survival space into the six-element finite compensation-motif family $\mathcal C=\{T,O,M,Q,P,D\}$ (temporal pulse, operator angle, mean variation, seven-point quadratic cancellation, pressure concentration, derivative-gate defect). C5 no longer asks which branch can still be split off, but instead asks whether these motifs, after record-window renormalization, can admit a compatible recurrent limit. The document first lays down a hard rule: Seregin's necessary condition proves that under a hypothetical blow-up, $\|u(t)\|_{L^3}\to\infty$ and $\|u(t)\|_{\dot H^{1/2}}\to\infty$, so within the singular-lineage rescaling, no uniform upper bound currently exists to which the standard Gallagher–Koch–Planchon critical-element compactness machinery could be applied — C5 explicitly forbids assuming compactness of the full critical field directly, and instead adopts metadata/probability-measure/defect-measure compactification. The record window is first given a unit-time renormalization ($s=(t-\tau_j)/L_j\in(0,1)$), but its relative viscous scale $\Theta_j^{time}=\nu\lambda_{q_j}^2L_j$ must additionally be retained — unit-time normalization is not the same as parabolic-time normalization. The six motifs are then each converted into genuinely compact objects: the middle-strain load becomes a probability measure on $\mathcal P([0,1])$; operator positive/negative growth becomes sub-probability measures, with a compensation offset $\beta_j^{op}=(P_j-N_j)/(P_j+N_j)\in(0,1]$ attached (guaranteed by C4-J's exact identity $P_j-N_j=\Delta E_{1,j}$); operator angle uses a bounded coordinate change so that even a ratio $r_\nu\to\infty$ still lands in a compact space; mean variation becomes a vector measure of bounded total variation; C4-J's seven-point Carathéodory witnesses become the compact space $\Delta_7\times(S^5)^7$; pressure concentration becomes a probability measure on the sphere; and the derivative defect uses a one-point compactification $\mathbb N_\infty=\mathbb N\cup\{\infty\}$, allowing the derivative order itself to diverge as a legitimate boundary value. The round's core result is the Compensation-Motif Sequential Compactness Theorem (Theorem 14.1, relying on finite-dimensional compact factors, weak compactness of probability measures on a compact metric space, weak-star compactness of bounded vector measures, and stitching together the finite discrete motif states one by one): every infinite C4-J record sequence has a subsequence along which every component converges, in its own topology, to a unified state $\Theta_\ast^{C5}$. The document immediately draws a sharp line around what this does not prove (§15): it does not prove that the rescaled field itself converges in $L^3$, $\dot H^{1/2}$, or any full critical topology, nor does it prove that this limiting state is actually produced by some genuine N-S limit field — it is only a necessary motif-compatibility limit state. C5-A.2 is the round's most important new no-go: an explicit construction proves that weak limits can erase microscopic pulse separation — cutting $[0,1]$ into rapidly alternating equal-length cells, with $m_j$ taking its value on even cells and $o_j$ on odd cells, so that $m_j(s)o_j(s)=0$ holds everywhere for every $j$ (completely disjoint at every finite scale), yet the weak limits of both converge to the very same Lebesgue measure — proving that weak-limit overlap is not the same as microscopic same-instant synchronization, which forces the next step to introduce temporal Young / two-scale defects. The document also provides a scale-dependent overlap spectrum $\mathfrak O_{j,n}$ (using a triangular kernel $K_n$) as a partial remedy, capable of preserving overlap information at each fixed scale $n$ without misreading it as genuine synchronization. The round closes with five explicit no-gos (§27): motif compactness does not imply field compactness; pairwise disjointness at every $j$ does not imply disjointness of the limit; equal limiting measures do not imply finite-scale simultaneous overlap; a seven-point cancellation coefficient tending to zero does not imply that seven spatial points are genuinely cancelling exactly; and a pressure limiting measure being an atomic measure does not imply that a genuine N-S singularity actually occurs. It formally hands off to C5-B, Temporal Young Defects and Pulse-Phase Compatibility, whose task is to recover, via a colored temporal Young measure, the microscopic phase information deliberately discarded here.