I have always found it odd how some databases conflate network connections with threads (or processes) of execution. If Postgres has an ideal number of concurrent workers it should instantiate that many and multiplex requests from various connections across them which has several benefits including allowing you to preemptively shed requests you know you won't get to in time and do other scheduling and QoS stuff.

I get it from a historical perspective where it would have been needless complexity to implement something else.

On an unrelated note it's pretty amazing what threads get conflated with in general.