feat(v4): the system hands no GONE on; (GONE-HOOK) goes (MESH step 7, increment 2)

Ruling 7 as amended: the kernel tells a node's going, so the system's
hand-on is not used.  (GONE-HOOK), (DOWN) and (RUN) go from quit.f18, and
COLD's hook-clearing from system.f18; the cell is free again.  The tests of
the hand-on go with it (mesh: the hand-on checks and the fourth disturbed
exchange; depth sweep: path 10, as before 529ba0fc).  A believed GONE going
no further is still checked.  test_host_unit.c's message for GONE down the
graph now says the kernel tells.  MESH.md: 9a.6 items 5 and 7, and step 7's
record of both increments.

Suite EXIT 0; depth sweep 230 runs, 0 failures; mkcapsule lint 0
violations.  Hosted and bare metal, all three ISAs: word_count=317
dict_hash=0xfec8c47cfc6f437f.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

docs/v4.0.0/MESH.md

@@ -1524,18 +1524,23 @@ Each test is written to fail first.
 4. **GONE down the graph.** A node of unit 2 waiting on a node of unit 1
    ends "Node gone" when Hera kills that node. A GONE from any node but a
    node's own centre is still not believed.
-5. **The new message path** (a centre passing a GONE on to the nodes it
+5. ~~**The new message path** (a centre passing a GONE on to the nodes it
    birthed): every depth of both stacks (`v4/tests/test_host_depth.c`);
    and the passing on disturbed at every engine step -- the centre
    removed, a child removed, the wire cut -- with every node left coming
-   to rest and answering (`v4/tests/test_host_mesh.c`'s way).
+   to rest and answering (`v4/tests/test_host_mesh.c`'s way).~~
+   **Withdrawn with ruling 7 (2026-10-16):** no node passes a GONE on, so
+   there is no new message path on a node. The kernel's telling is tested
+   in C at both widths (`v4/tests/test_manage.c`) and in the fabric
+   (`v4/tests/test_host_unit.c`, item 4).
 6. **The kernel's side.** The registry and who may ask, tested in C at
    both cell widths.
 7. **Nothing else changes.** `make -C v4 test test-32 sanitize
    hosted-check`; `mkcapsule --lint capsules/`; the three QEMU boots and
-   `make -C v4 test-32`, as `.claude/CLAUDE.md` says. The system changes
-   (`(GONE-HOOK)`), so the products' `dict_hash` changes; it must be the
-   same on all three ISAs (ruled 2026-10-16, planning).
+   `make -C v4 test-32`, as `.claude/CLAUDE.md` says. The system changes,
+   so the products' `dict_hash` changes; it must be the same on all three
+   ISAs (ruled 2026-10-16, planning). (The change was first `(GONE-HOOK)`;
+   with ruling 7 it is that word's going.)
 
 **A limit kept, not widened:** `test_host_unit.c` is not run at 32 bits;
 POST's expected results are 64-bit (step 5).
@@ -2114,6 +2119,31 @@ Each is tested, committed and pushed before the next.
      (riscv64): POST 538 of 538, `word_count=318`,
      `dict_hash=0x248b4edc47f4357e` on all six systems (the system changed,
      so the hash did); `make -C v4 test-32`, 0 failures.
+   - **Amended 2026-10-20, ruling 7 (the kernel tells a node's going).**
+     Two increments, each accepted on its own.
+     *Increment 1* (5707b07a): `manage.c`'s `KILL` puts a GONE about the
+     node on the wire of every live node from its parent, as if its parent
+     had sent it; `centre.4th` loses `GONE-DOWN` and its table of the
+     nodes it birthed; `hera.4th`'s `KILL` is the kernel's kill and
+     `NO-ROUTE`. Accepted in 9138ebd8: `logs/20261020-084230` (amd64),
+     `-084520` (aarch64), `-084933` (riscv64), POST 538 of 538,
+     `dict_hash=0x248b4edc47f4357e` on all three.
+     *Increment 2*: the system's hand-on goes -- `(GONE-HOOK)`, `(DOWN)`
+     and `(RUN)` from `quit.f18`, `COLD`'s hook-clearing from
+     `system.f18`; the cell is free again (`v4/tests/host_map.h`). The
+     tests of the hand-on go with it: the mesh test's hand-on checks and
+     its fourth disturbed exchange, and the depth sweep's path "AWAIT
+     ended by a GONE that is handed on", which is again as before
+     529ba0fc. What stays: a GONE believed from a node's centre goes no
+     further (`v4/tests/test_host_mesh.c`, 244 checks); GONE down the graph
+     ends a wait in the second unit (`v4/tests/test_host_unit.c`, 126
+     checks); the kernel's side (`v4/tests/test_manage.c`, 52 checks, both
+     widths). Verified: `make -C v4 test test-32 sanitize hosted-check`
+     (depth sweep 230 runs, 0 failures); `mkcapsule --lint capsules/`, 0
+     violations; three bare-metal boots, `logs/20261020-125114` (amd64),
+     `-125411` (aarch64), `-125820` (riscv64): POST 538 of 538, `PARITY:OK`,
+     `word_count=317`, `dict_hash=0xfec8c47cfc6f437f` on all six systems;
+     `make -C v4 test-32`, 0 failures.
 8. **The hosted product is the five nodes**, on three ISAs: acceptance 1
    to 5.
 9. **The same on bare metal:** acceptance 6.