Fix new pg_publication_tables query.

The addition of published column names forgot to filter on attisdropped,
leading to cases where you could see "........pg.dropped.1........"
or the like as a reportedly-published column.

While we're here, rewrite the new subquery to get a more efficient plan
for it.

Hou Zhijie, per report from Jaime Casanova.  Back-patch to v15 where
the bug was introduced.  (Sadly, this means we need a post-beta4
catversion bump before beta4 has even hit the streets.  I see no
good alternative though.)

Discussion: https://postgr.es/m/Yxa1SU4nH2HfN3/i@ahch-to

src/backend/catalog/system_views.sql

@@ -370,11 +370,10 @@ CREATE VIEW pg_publication_tables AS
         N.nspname AS schemaname,
         C.relname AS tablename,
         ( SELECT array_agg(a.attname ORDER BY a.attnum)
-          FROM unnest(CASE WHEN GPT.attrs IS NOT NULL THEN GPT.attrs
-                      ELSE (SELECT array_agg(g) FROM generate_series(1, C.relnatts) g)
-                      END) k
-               JOIN pg_attribute a
-                    ON (a.attrelid = GPT.relid AND a.attnum = k)
+          FROM pg_attribute a
+          WHERE a.attrelid = GPT.relid AND a.attnum > 0 AND
+                NOT a.attisdropped AND
+                (a.attnum = ANY(GPT.attrs) OR GPT.attrs IS NULL)
         ) AS attnames,
         pg_get_expr(GPT.qual, GPT.relid) AS rowfilter
     FROM pg_publication P,