> And if they did -- if something needed to join ID values to display names, presumably those would sit in a database, not a different stream?

At a high level the push-instead-of-pull benefit here is "you don't have to query the ID values to get the display names every time" which will reduce your latency. (You can cache but then you might get into invalidation issues and start thinking "why not just send the updates directly to my cache instead")

There's also a less cacheable version where both sides are updating more frequently and you have logic like "if X=1 and Y=2 do Z."

For small enough batches streaming and micro-batching do often end up very similar.