[K3] Cut the source-bound compiler at the active carrier bridge

[scottdhughes]

Stack and review boundary

This is a success-condition-B K3 route-cut packet stacked on ready PR #1152 at exact head ed4877cce5f227f33311fa93f5ff5e5f4150ae63.

The net review boundary is:

ed4877cce5f227f33311fa93f5ff5e5f4150ae63..d7f0fd9370b3c13ff93293f08e03cadddb59b921

PR #1152 is still open and its artifacts are absent from main, so GitHub shows the dependency stack until #1152 integrates. After integration, replay only this one packet commit onto the integration head.

Result

The maximal source-bound K3 attack terminates rigorously in route cut B, not a K3 payment.

The packet is scoped only to the declared active residual slice Z_BC^(m=2,r=4). It does not claim a source-bound membership predicate for that slice or an exhaustive reduction from all Z_BC.

It identifies the first missing implication:

declared-slice bad slope
  -> total canonical selector into a supplied actual record/component
  -> reconstruction preserving line, support, slope, owner, and chronology
  -> exact finite projection-fiber bound
  -> raw workboard
  -> distinct-affine-slope payment

A two-sided inverse would suffice but is stronger than necessary. The pinned source-pencil theorem works per supplied actual endpoint record and explicitly distinguishes its parameter line from the evaluation carrier.

Exact workboard and route cuts

• Replays the raw 433-1b -> O0a exclusion at public-DAG commit 8df0903391a2: 15 cells, 1,575 labels, 25,200 signed systems, zero survivors.
• Reconstructs the exact thirteen-route table: two raw-empty routes and eleven unresolved raw-workboard routes.
• Keeps both raw-empty routes at surviving_raw_systems=0 but distinct_affine_slope_payment=null.
• Reconciles [K3] Cut the 433-1b cell-11 guard transplant #1155 as a valid guard-transplant route cut with its exact surviving point; it is not a payment.
• Cuts the proposed 433-1b -> O0b / closed-O0a transport only in the frozen natural record/role/gauge/typed-incidence class:
  • repeated BC has parallel-product ratio +1 versus -1;
  • split BC has zero isomorphisms under all six typed outside-role permutations.
• Makes no claim against arbitrary orientation-changing algebraic maps.
• Records 11304 only as a conditional candidate abstract-label-orbit workload—not a census, elimination, representative count, or payment.
• Records eight unresolved candidate source-cover row/passport combinations—not terminal cells.

Exact ledger output

All requested K3 quantities remain deliberately undefined:

U_remaining       = null
U_positive        = null
U_sourcecover     = null
U_K3              = null
U_K3_allocation   = null
signed_slack      = null
ledger movement   = 0

The tangent manifest declares U_paid=981104, but this packet does not bank it: the manifest expects grande_finale.tex blob 8a5d979..., while #1152 contains 6b21d6e.... Both tangent-verifier modes fail exactly at that stale proof-source gate. The derived reserve is manifest arithmetic, not a K3 allocation.

Validation

• primary Python replay, normal and -O: PASS;
• 13 routes, 2 raw-zero, 11 open, 32/32 shipped hostile mutations rejected;
• 23/23 direct source bindings checked;
• independent Sage replay: PASS, including all 10,395 involutions;
• independent FLINT replay: PASS;
• raw aggregate verifier and all three raw node verifier/audit pairs: PASS;
• [K3] Cut the 433-1b cell-11 guard transplant #1155 normal and -O: PASS, OPEN_GAP, 16/16 mutations rejected;
• Draft 2020-12 schema and canonical manifest checks: PASS;
• exact tree on the published branch matches the locally reviewed tree;
• fresh independent proof review: GREEN / PR-ready for success condition B;
• fresh independent certificate/custody review: GREEN / PR-ready, plus 41 additional promotion mutations rejected.

The tangent verifier's expected stale-source failure is a quarantined diagnostic, not a passing test.

Upstream overlap and nonclaims

A refreshed audit of all 23 open K3 PRs found no duplicate packet path or active-slice selector/projection theorem. #1130-#1132 explicitly leave the relevant bridge open; #1155 is a distinct guard-transplant cut; draft #1156 repairs a different scalar-locator source pin and does not repair the tangent adapter.

This PR does not close K3, the KoalaBear row, all of Z_BC, any of the eleven unresolved raw routes, source-cover, or the stale tangent payment. It moves no ledger value.