If simple checks like "ssh" are mixed with agent checks, the host availability of hosts that are unreachable becomes unstable.

Since the only criteria for enable_host seems to be the existence of at least one item that can be fetched, the simple check will cause the host availability to be set to HOST_AVAILABLE_TRUE, even if the check is negative (e.g. because the ping or ssh check failed to reach the host).
On the other hand, the agent checks will cause the availability to be set back to HOST_AVAILABLE_FALSE after CONFIG_UNREACHABLE_PERIOD.

This leads to continuous flapping of the host availability and thus the "status" item will also continuously flap.