[K3] Audit actual-record dimension shift before Rec_2_4

[scottdhughes]

Stack and review boundary

This is a ready, zero-payment K3 audit stacked on #1158 at exact head 5145fc8e0abca6325b8226294cdc2661e0891dcc.

Review only:

5145fc8e0abca6325b8226294cdc2661e0891dcc..e603e0cedc5220ec2f29bd53836e732e3ec14934

Upstream main remains 93fba1be3f3299b0ba4708d88715377bbb656e45; #1157 and #1158 remain open and ready. Their artifacts were not manually integrated: that was checked by pinned content and hashes, not only PR metadata.

Exact result

The packet constructs a deployed KoalaBear pole-line MCA-bad slope alpha with exact size-m support and proves

d1^(k) = d1^(k+1) = 67473.

The same actual record therefore has these exact numerical profiles:

• deployed K=k: boundary, (d1,d2,m',deg B)=(67473,981104,1116048,<=0);
• finite-calibration K=k+1: first interior, (67473,981103,1116049,<=1).

This is an exact dimension-sensitivity theorem for one actual received-line/slope record. The proof uses a direct interpolation-lattice basis and root-count argument; toy-field ranks are regression controls only.

Semantic verdict

This is not success A or B from the post-#1158 attack.

The pinned v4 sources do not establish that this actual record belongs to Q, balanced core, U_new, a primitive endpoint, or any #1157 route. In particular, no owner collision or source-level non-definability theorem is claimed.

The first missing bridge is isolated as SEM-QBC: a source theorem connecting the deployed actual-slope/witness chronology to the Q/BC priority map and then to the endpoint relation consumed by the public-DAG selector. Until that theorem is supplied, Rec_2_4, its fibers, the all-Z_BC complement fence, and K3 payment remain open.

Ledger effect

The certificate binds:

• U_paid = 981104;
• U_Q, U_list-int, U_ext, U_new, and every K3 total: unassigned;
• ledger movement: 0.

No multiplicity or allocation is imported from endpoint labels, passports, gauges, or another partition.

Reconciliation

The packet explicitly reconciles #1130, #1132, #1139, #1143, #1152, #1155, #1156, #1157, and #1158. It preserves #1155 as a guard-transplant route cut and treats #1157/#1158 as the current K3 boundary. It neither reopens raw cell-11 elimination nor duplicates #1156's scalar-locator repair.

Verification

Frozen canonical payload:

75ee17e4de50c5a4a0acd360c9d94ba00748875527390212823631c4585b4970

Passed:

• primary Python verifier in normal and optimized modes;
• five local and six public-DAG source pins;
• 62 semantic and 3 parser mutations in both modes;
• missing-DAG-root fail-closed check;
• draft-2020-12 schema validation;
• Sage exact finite-field replay;
• FLINT independent polynomial/root and matrix replay;
• Wolfram independent symbolic identities and integer totals;
• exact huge-binomial fingerprint and canonical JSON/hash checks;
• [K3] Cut the source-bound compiler at the active carrier bridge #1157/[K3] Cut the active-slice bridge at direct carrier inference #1158 predecessor, tamper, Sage, FLINT, and tangent checks;
• git diff --check.

Independent mathematics review is GREEN for the narrow theorem and RED for any K3 closure/payment claim. Independent certificate review is GREEN, including 309 resealed one-leaf mutations with zero accepted and strict validation of all 24 object rules.

Maximal next theorem

Construct SEM-QBC from the pinned source chronology, then prove the actual-record Rec_2_4 selector/projection, exact fibers, all-Z_BC complement fence, and complete K3 payment. This PR deliberately moves no ledger mass before that bridge exists.