Record shell-emission slice-2 landing in authority carriers (Wave C4 closeout: keystone already merged as #6572/#6585/#6598) - #6948

What this is

Wave C4's brief — route fleet_converge_emit's ~275 lines of Doc-literal bash through the EmitArtifactThenThinRun transport — describes a tree state that no longer exists. The keystone is already merged on main:

• Shell→dag P4 KEYSTONE: ConvergePlan + EmitArtifactThenThinRun + converge_apply #6572 — ConvergePlan effect + EmitArtifactThenThinRun transport + converge_apply in-process knob interpreter (G1+G2)
• Shell→dag P4-G3 completion: fleet_converge steady-state arms move IN-PROCESS via the landed converge_apply/ConvergePlan (#6572); committed .github/fleet-converge.sh golden SHRINKS from 323 lines to fresh-standup bootstrap + thin gunbc-converge invocation; bootstrap arm re-emitted as ONE intent tree #6585 — fleet_converge steady-state arms moved in-process; committed .github/fleet-converge.sh golden shrunk 323 → 21 lines (G3)
• Shell→dag: wire the gunbc converge CLI receiver — stage0 Subcommand gains converge --host <h>, dispatching to the landed converge_apply/ConvergePlan in-process interpreter (#6572) with typed receipts on the byte-locked receipt grammar; green by EXECUTION on a fixture host + discriminating RED (unkno #6598 — gunbc converge --host <h> CLI receiver with typed receipts
• FLAGs 2a(i)/2b/2c signed/discharged by the operator 2026-07-14 (census §2)

Receipts re-verified by execution 2026-07-20 in this session's worktree: fleet_converge_emit_bootstrap_holds / fleet_converge_emit_thin_run_holds / fleet_converge_emit_policy_holds / fleet_converge_emit_holds and fleet_converge_apply_holds all PASS — including the drift gate vs the committed golden and its one-byte perturb RED.

Why a PR at all

The stale "~275 lines / 12+ fn defs" framing has now caused two worker dispatches onto finished work (slice 0 on 2026-07-16, slice 2 on 2026-07-20 — this session). Mirroring #6878 (the Phase-0/1 landing-record precedent: "update design doc, roadmap, and open-thread status so workers are not re-dispatched onto finished work"), this PR records the landing in the authority carriers:

• roadmap_authority.dag 6-shell-slice2 → MERGED, operator sign-off PENDING with landing PRs + receipt date. done stays false — the signed(operator) attestation is the operator's alone (mirrors 6-shell-intent-phase1's shape; no forged sign).
• plans/shell_emission_model.dag §1 status + slice-2 row → merged/receipt state, with a dispatch note.
• docs/plans/shell-to-dag-residual-census-and-arc-completion.md (hand-authored, no carrier): category-C fleet_converge_emit row marked dissolved with receipts; §2 goal marked achieved (sign-off pending); sequence steps 1–2 landed, step 3 annotated still-open (fleet_converge_steady_state_doc_projection_dissolution_trigger remains a Scaffold binding fleet_converge_thin_invocation — dissolves when the thin invocation becomes typed argv dispatch); P2 consumer slice marked landed (Shell→dag P2b consumer slice (successor to archived bold-pike-629; plan agreed on-record): (1) route converge_gunbc_pinned_tree fresh_standup fragment through emit(intent,Bash) via procedure/cmdsubst_assign/let_assign ONLY (steady-state functions untouched — P4 lane owns those); (2) byte-identity wi #6573); critical-path summary flipped from "P4's execution" to P5/P6 + the operator sign.
• ROADMAP.md / docs/plans/shell-emission-model.md / .github/live-deploy-srv1-apply.sh regenerated via generated_artifact_gate main_wet (the live-deploy script embeds the roadmap heredoc, so it regenerates with the carrier — same as Record shell→intent Phase 0/1 landings in authority carriers (supersedes #6852) #6878).

Verification

• generated_artifact_gate main (drift gate + perturb RED over all committed artifacts): ExitSuccess
• roadmap_authority_witnesses, roadmap_page_keystone_holds: PASS
• fleet-converge witness suites (above): PASS

Operator: the remaining slice-2 action is yours — sign 6-shell-slice2 (done: true + sign(operator)) when satisfied; this PR deliberately does not.

🤖 Generated with Claude Code

Also fixes: #6947 host_standup spine-witness regression (masked on main)

ba14c974eb fixes a live main regression inherited via the main merge: #6947 rewrote the prefix:os-install-actuated DeclaredGap interim_realization and dropped every extdeps. citation, reding declared_gap_steps_carry_interim_realization_authority (host_standup_spine_witness_test.dag). Main's CI cannot see it — it fails earlier at regen_verify_gate_passes (batch 1, #6848 fallout), so the discovery corpus never runs there. Fix: cite extdeps.bmc.os_install_actuator_toolchain in the spine row (the same interim authority #6947's own gap-ledger row kept) rather than weakening the witness; verified red-before/green-after, 16/16 spine witnesses pass. Confirmed correct by calm-ferret-849 (C3 reviewer). Merging this PR un-masks and clears that regression.