test: report a bare docs/adr/NNNN citation as unresolved

[lesnik512]

tests/test_adr_citations.py only matched docs/adr/NNNN-<slug>.md, so a citation written as a bare docs/adr/NNNN was invisible to it. The pattern now also captures the bare form, which names no file and so falls out of the existing is_file() check as unresolved. Same change as modern-python/httpware#129; the file stays byte-identical across the org.

Also: the lint job was red on main

CI runs uv lock --upgrade, and ty 0.0.82 (2026-09-17) started reporting OutboxParamsStorage as abstract: DefaultLoggerStorage inherits register_subscriber from the LoggerParamsStorage protocol without implementing it. At runtime the protocol's ... body is inherited, so nothing broke; the gap was type-level only.

Fixed the way faststream's own RedisParamsStorage / KafkaParamsStorage do it: register_subscriber records the longest queue name and get_logger uses it as the column width, so the access log stays aligned for queue names longer than the previous fixed 7. Test first (it failed on the old fixed width), then the implementation. Checked with both the locked ty 0.0.81 and 0.0.82.

[github-actions]

Benchmark gate

✅ gate passed

scenario	msg/s	delete/msg	WALrec/msg	WALB/msg	fpi	upd	del	dead_tup
consumer/w1/b10	990	1.000	6.82	909	242	5000	5000	10000
consumer/w1/b100	929	1.000	6.87	916	243	5000	5000	10000
consumer/w2/b10	918	1.000	6.79	957	242	5000	5000	10000
consumer/w2/b100	1079	1.000	6.78	988	265	5000	5000	10000
consumer/w4/b10	635	1.000	6.66	916	244	5000	5000	10000
consumer/w4/b100	1526	1.000	6.84	1027	243	5000	5000	10000
consumer/w1/b100/tfbs100	7314	0.010	6.07	1116	243	5000	5000	10000
producer/w1/b100	3043	0.000	3.04	584	0	0	0	0

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