Replace the metadata-only subplan spawn with real child Plan domain object
creation. SubplanService.spawn() now creates full Plan instances with proper
PlanIdentity (linking parent_plan_id and root_plan_id), sets them to
PlanPhase.STRATEGIZE / ProcessingState.QUEUED, and returns them in the
SpawnResult.child_plans list for downstream lifecycle orchestration.
Key changes:
- SubplanService.spawn() now creates Plan domain objects for each spawn entry
with complete PlanIdentity linking, inheriting the parent plan's namespace,
actors, project links, definition_of_done, and access settings.
- SpawnResult dataclass extended with child_plans: list[Plan] field to carry
the created child plans alongside the existing metadata and statuses.
- Parent plan's subplan_statuses list is updated with SubplanStatus entries
tracking each child's lifecycle state.
- Fixed Pyright type errors: added missing definition_of_done, reusable,
read_only, and server parameters to Plan and NamespacedName constructors.
- Removed @tdd_expected_fail tags from TDD test files since the bug is fixed.
- Added Behave scenarios and Robot integration tests verifying child plan
creation, lifecycle phase/state, and parent tracking.
ISSUES CLOSED: #823
