[lesnik512]

Summary

The broker.publish docstring still said that re-publishing a timer_id "is a no-op (returns None)". That was true before timer_conflict existed. Under timer_conflict="extend" it is wrong, because an unleased timer is updated and the call returns its id. The next paragraph of the docstring already describes both modes correctly, so the timer_id sentence now points to it instead of contradicting it.

OutboxPublisher.publish and OutboxResponse were updated in #254 and need no change.

Checklist

• ruff format --check and ruff check pass on broker.py; tests/test_unit.py and tests/test_fake.py pass (461).

[github-actions]

Benchmark gate

✅ gate passed

scenario	msg/s	delete/msg	WALrec/msg	WALB/msg	fpi	upd	del	dead_tup
consumer/w1/b10	1483	1.000	6.72	921	242	5000	5000	10000
consumer/w1/b100	1622	1.000	7.11	1154	243	5000	5000	10000
consumer/w2/b10	1831	1.000	6.72	921	242	5000	5000	10000
consumer/w2/b100	2165	1.000	7.11	1154	243	5000	5000	10000
consumer/w4/b10	1809	1.000	6.75	951	242	5000	5000	10000
consumer/w4/b100	2330	1.000	7.11	1154	243	5000	5000	10000
consumer/w1/b100/tfbs100	7460	0.010	6.11	1130	243	5000	5000	10000
consumer/w1/b100/interval1-10	1628	1.000	7.17	1196	265	5000	5000	10000
producer/w1/b100	3273	0.000	3.04	597	1	0	0	0
producer/w1/b100/ppt1	1085	0.000	4.04	635	0	0	0	0

Gated (fails the build): delete_calls + tuple counters (upd/del/ins) + the producer's insert_calls, exact; select_calls and other_calls within +2; wal_records within a 10% band; fetch_calls within 2x. msg/s, WAL bytes and total calls are informational (timing/FPI noise).