It all depends, right? 😊
We run a collapsed cluster for everything, our edges and our management run in the same cluster as our workloads. This has been going fine for years, but we're starting to get to a point where we want to take our management to a separate cluster.
The main driver for this decision has been cost and reduced complexity. We use a stretched cluster and only run 1 edge node in every DC. No concerns whatsoever from a performance point of view, as we always keep our cluster below 50% resource utilization.