Unfortunately, the only way to change column order is to either recreate the table, or add new columns and move data.
[1]: https://www.postgresql.org/message-id/flat/20220628083230.c4...
I'm about to move off from PostgreSQL back to MySQL primarily for this reason.
After a period of development, the ordering becomes a mess by keep adding columns to the end with no logical grouping and it's a pain to view the data that way.
Some cool optimizations are possible if the size of each row doesn't change. * Rows could be done in parallel. * The simplest method would be to copy the row to a scratch buffer, and then copy over data to their final position. * Furthermore, with some fancy permutation, you only need a register of scratch space. Although, with more random memory access, that could result in worse performance.