fix(kanban): notify tool-created tasks

[eloklam]

Summary

Kanban tasks created through the tool path could finish or block silently because they never received gateway notification subscriptions.

Slash-command-created tasks already call add_notify_sub(...), but orchestrator/worker fan-out goes through tools/kanban_tools.py::_handle_create, which previously only created the task row. That left kanban_notify_subs empty, so the gateway notifier had no destination when the task reached a terminal state.

This PR adds notification setup to tool-created tasks and preserves subscriptions across child-task fan-out.

Changes

• Add _ensure_notify_subscriptions(...) in tools/kanban_tools.py.
• Copy existing parent kanban_notify_subs rows onto newly created child tasks.
• If no parent subscription exists, read the active gateway session via gateway.session_context.get_session_env(...) and subscribe the origin platform/chat/thread.
• Lazy-import gateway session context so normal CLI/test imports stay safe.
• Preserve CLI/no-session behavior: if no gateway session is present, no subscription is created.
• Add regression tests for origin chat auto-subscription, child task inheritance, and no phantom subscriptions without parent/session context.
• Add AUTHOR_MAP entry for contributor attribution check (scripts/release.py).

Validation

• scripts/run_tests.sh tests/tools/test_kanban_tools.py::test_create_auto_subscribes_origin_in_gateway_session tests/tools/test_kanban_tools.py::test_create_child_inherits_parent_notify_subscriptions tests/tools/test_kanban_tools.py::test_create_no_subscription_without_parent_or_session tests/hermes_cli/test_kanban_notify.py -q
  • 13 passed in 12.56s

Notes

• No linked issue.
• No schema migration; this reuses the existing kanban_notify_subs table and notifier flow.

[alt-glitch]

Related to #21523, #24307 (duplicate of merged-then-reverted #19718/#19721), and tracking issue #19479. The merge/revert cycle on #19718 left this gap unfixed.

[eloklam]

Related to #21523, #24307 (duplicate of merged-then-reverted #19718/#19721), and tracking issue #19479. The merge/revert cycle on #19718 left this gap unfixed.

Aint no way this is P3 bro

[teknium1]

Merged via PR #28433 (cherry-picked onto current main with your authorship preserved via rebase-merge — commit 9d9f316). Thanks for the contribution!