For instance, I used extensively tps throttling, latencies and timeouts measures when developping and testing the checkpointer sorting & throttling patch.
I have to admit, I've found tps throttling and latency measurement useful when working with logical replication. It's really handy to find a stable, sustainable throughput on master at which a replica can keep up.
PostgreSQL is about more than raw TPS. Users care about latency. Things we change affect latency. New index tricks like batching updates; sync commit changes for standby consistency, etc.
That's not a reason to throw anything and everything into pgbench. But there's value to more than measuring raw tps.