Thanks for sharing, some of the recommendations are a bit odd…
CREATE INDEX index_topics_on_category_id ON public.topics USING btree (category_id)
CREATE INDEX index_topics_on_category_id_and_id ON public.topics USING btree (category_id, id)
CREATE INDEX index_topics_on_id_and_category_id ON public.topics USING btree (category_id, id)
This triplicate dupe index it is suggesting has extremely low selectivity, the trouble with indexes like that is that the planner correctly tends to ignore them. If there are only 10 different values you need to mix it up with something else like say topic_id
As to underused indexes, about 6 months ago we ran some stats on the database looking for indexes we did not touch in our infrastructure, we did prune a few but this topic is a good reminder for us to run through this process again.
Thanks for posting this.