AWS Glue crawler does not detect array column when the source database is RDS(Postgres)

Question

I have source tables in Postgres. When I run the crawler, the tables which have a column as an array in Postgres are detected as string when the crawler runs.

I have an option to edit the data catalog manually and change the column from string to array.

But if I rerun the crawler it's updating back to a string.

Below is the table in RDS and its schema

 Table "public.samples"
    Column     |              Type              | Collation | Nullable |                Default
---------------+--------------------------------+-----------+----------+---------------------------------------
 id            | bigint                         |           | not null | nextval('samples_id_seq'::regclass)
 title         | text                           |           | not null |
 created_at    | timestamp(6) without time zone |           | not null |
 updated_at    | timestamp(6) without time zone |           | not null |
 topics        | text[]                         |           |          | '{}'::text[]
 sections      | text[]                         |           |          | '{}'::text[]

Below is the schema of the table which crawler detected