Brummel
5e401a3520
Records the two emission seams (pattern-binder at arm close, Own-param at fn return), why they fit one iter (uniform emission shape), the explicit dynamic-tag partial-drop fallback to shallow ailang_rc_dec when moved_slots is non-empty for a dynamic-runtime-ctor binder, the side effect on reuse_as_demo's sum_list (correctness improvement, stdout unchanged, existing assertions still hold), and the closing of 18d.3's regression. Notes that 18d.3 + 18d.4 together close the move-aware-pattern story for the static case; dynamic-tag partial-drop is the remaining hygiene gap and is structurally orthogonal (would exist even without moves). 18e (worklist) may be the natural place to also close the dynamic-tag fallback.