v1 closeout wave 2: merge #13583, #13582, #13615, #13609, gentle-dove-36, #13108 + regen

[gunbai-bot]

v1 closeout wave 2: conflict resolutions + one regen round

Base: integration/v1-closeout (#13641). Not for merge/enqueue by this session.

Merges (plain merge commits, origin/<branch> by name)

1. One directory authority for dashboard instances (D14 OwnedDirectory slice) #13583 deploy/dashboard-owned-directory-authority: live_deploy emit/member_observe/spec take One directory authority for dashboard instances (D14 OwnedDirectory slice) #13583's host_directories directory authority; emit_test keeps semver: route the version scheme's identity compare through the cited precedence rule #13412's semver_version_parse call.
2. Delete the v2 packrat parse memo carrier (R2, stacked on #13577) #13582 parse-memo-delete then Delete unused memo and preparation carriers (native-memory-inert-carriers) #13615 inert-memo-carriers: floor_grandfathered_roster both blocks kept; floor_preparation_witness_test.dag stays DELETED; 02_parse.dag modify/delete resolved to theirs; doc_graph_roots.dag kept HEAD.
3. Derive ledger roster modules without mutating source roots #13609 seed/derived-rosters-in-memory: local_repo_wet_terminal.dag claims from both sides.
4. integration/gentle-dove-36: the three T?==T witnesses resolved to one form each; generated mirrors regenerated (below).
5. admit_callers: restore the native-route caller-admission wall (DP-M6 item 2) #13108 eager-cat-664-admit-callers: 04_infer.dag / body_lowering_fold.dag import unions.

#13574 reverted (e2b6cb8)

Operator decision (relayed by neat-wolf-604): #13574 (04_infer interim equality wall, mirror, native_emission_controls/witness rows) is reverted out of this snapshot; it is dropped here, not landed. It carries an open REQUEST_CHANGES (review 78262), not addressed here. #13595 left out.

Regen

Per the 2026-10-04 ruling one emission round is the regen. The first seed would not build because bootstrap mirrors lacked rt_host_budget; two provisional bootstrap commits seeded it. Round 0 (BuildBuddy 81e24bb4) and round 1 (build failed on that gap) were superseded by round 2 (6 min run, 20 files), committed as ed8cc53: v1_rt.rs is now emitted from runtime_rust.dag rt_host_budget (HostBudgetJoin*/HostBudgetCgroupV1 included); no hand-appended lines remain. first_generation_equal was false in every round (not required by the ruling).

Gate on ed8cc53 (BuildBuddy de8e52a0, one dispatch)

cargo build --release -p v1-compiler --bin gunbc exit 0; gunbc test //gunbc/instruments:v2-native-cli exit 0; cargo clippy --all-targets -- -D warnings exit 0.

Not run here: generated_artifact_gate main_wet (docs/design-rung-drops.md and fleet-converge.yml were not in the regen output; CI's build lane is the check).

🤖 Generated with Claude Code

[gunbai-bot]

Reviews 78374 / 78378: addressed at ed8cc53. The hand-appended rt_host_budget lines were bootstrap only; the round-2 regen emitted v1_rt.rs from runtime_rust.dag with that section, so the authority does emit it and the appended lines are gone. Per the 2026-10-04 ruling one round is the regen. #13574 revert is an operator decision (stated in the body). — sent from crisp-ant-863