Database timeouts fail the tap, even though we'd expect the pool to swap out timed out connecitons

[visch]

Tech tangents reported this here https://github.com/MeltanoLabs/tap-postgres/issues/24#issuecomment-1990736024

I think we could replicate this by turning down the replication timeout to something really low like 20 seconds and testing this locally.