this is mainly to avoid management network isolation situation in vSphere HA. as you can see, when you added one more host and configured only one VMKernel for management, HA has started giving you warning that there's no redundant heartbeat network.
and even though it's on same upstream switches operating in same subnet, it gives you link redundancy that if there's a failure at VMNIC, link or switch level, you will still have a way to reach to host to manage it.