Can't dump new-style sequences independently from their tables.

Quick example:

xof=# CREATE TABLE t1 (id SERIAL PRIMARY KEY);
CREATE TABLE
xof=# CREATE TABLE t2 (id INTEGER PRIMARY KEY GENERATED ALWAYS AS IDENTITY);
CREATE TABLE
xof=# \d+
List of relations
Schema | Name | Type | Owner | Persistence | Access method | Size | Description
--------+-----------+----------+-------+-------------+---------------+------------+-------------
public | t1 | table | xof | permanent | heap | 0 bytes |
public | t1_id_seq | sequence | xof | permanent | | 8192 bytes |
public | t2 | table | xof | permanent | heap | 0 bytes |
public | t2_id_seq | sequence | xof | permanent | | 8192 bytes |
(4 rows)

Swift:~ xof$ pg_dump -t 't1_id_seq' | fgrep 'setval'
SELECT pg_catalog.setval('public.t1_id_seq', 1, false);
Swift:~ xof$ pg_dump -t 't2_id_seq' | fgrep 'setval'
Swift:~ xof$ pg_dump -t 't2' | fgrep 'setval'
SELECT pg_catalog.setval('public.t2_id_seq', 1, false);
Swift:~ xof$

So, you can dump a sequence created with SERIAL independently from the table it is owned by, but not a sequence created by GENERATED ALWAYS AS IDENTITY; you need to dump the owning table. It's easily worked around, but I'm curious why that is.