Seems like bulk loading might be a good use case here.
It’s a few years ago so details are blurry. We got daily snapshots and on ingest we created an unlogged temporary table then when complete used INSERT FROM to the destination table if I remember right.
As long as the application handles failure on bulk ingest and you copy the data to a logged table or enable logging on the table once ingested it worked really well.
The initial load might be faster, but you’ll end up repaying that time saving when you turn on logging back on, and Postgres is then forced to read the contents of the table and push it into the WAL anyway. Thereby throwing away any saving you originally made by avoiding the WAL with an unlogged table.
If you’re willing to throw away that functionality, then maybe you could avoid writing the rows to the WAL completely (but it seems Postgres doesn’t support that). But personally I would always be a little suspect of data stored in Postgres that never transited the WAL. It’s a core part of how Postgres provides it durability guarantees, and lots of tools and functionality assume that everything worth protecting gets written to it at least once.