Unless I'm overlooking something obvious one result has 500 000 rows the other 7 000.
You are right, it wasn't. I have 2 datasets, one containing 12 mil entries and the other 14 mil entries. I accidentally used the one with 12 mil entries in that table which actually only contains 7000~ entries for that sql query.
For now I have tested the 12.6 Postgres with default values and it finished in 12 seconds. I'll do some thorough testing and let you know once I finish, sorry for the confusion.
Are you sure you're using the same data det ?
Unless I'm overlooking something obvious one result has 500 000 rows the other 7 000.