It's a rare issue, but it should be addressed.

In my local database, I have found about 27 occurrences of an 'ARRAY' type in columns in uploaded PostGIS layers.

SELECT
    t.table_name,
    t.column_name,
    t.data_type,
    t.udt_name,
    s.id AS layer_name
FROM
    information_schema.columns t
JOIN
    mx_sources s
    ON t.table_name = s.id
WHERE
    t.udt_name LIKE '%_array%' -- Filter for array types
    OR t.data_type = 'ARRAY'   -- This checks for array types explicitly
ORDER BY
    t.table_name,
    t.column_name;

The proposed solution is to work upstream by checking each uploaded table and layer to see if any columns have been converted into PostgreSQL arrays. Then, we could write a post-process script to convert those columns into the JSONB type, which is supported by the table editor. Or just support the ARRAY type in the table editor. To be tested..

Another option would be to use a trigger on table creation, but that would affect all future tables, not only non-spatial mapx tables and PostGIS layers (mx_sources).