Disabling only stops the Hubitat driver code from executing. The original intent of this feature was to troubleshoot problematic user driver code: rather than removing all such devices or changing all devices using a specific driver, you could just disable devices that run it. If you have Z-Wave association, a secondary controller, or any other method to control the device, those should still work. I haven't tested this, but apparently disabling also sufficient to stop it from participating in a Z-Wave repair (or at least to stop anything from this device about appearing in the logs).
If the device is a repeater, it should continue to function as such; that is handled at the device/firmware level and is not something Hubitat can directly control. Personally, if I were not using a Z-Wave device for an extended period of time, I'd un-pair (exclude) it from my network and add it back when I planned to use it. That being said, Z-Wave Plus has decent "self healing," so if it sees a repeater isn't online for a while, devices should eventually realize that and stop using it. I've seen reports of people not removing devices like this and not noticing any problems. "Classic" Z-Wave devices may need a repair since they are not as capable on their own in this regard.
If anyone knows anything about Hubitat's Z-Wave implementation specifically (or what the spec might say, not that it necessarily matters since the C-5 isn't certified), I haven't seen that spelled out by staff or knowledgeable community members anywhere. My guess is that nodes that fail to respond after "a while" will get put on a "failed nodes" list until the hub/controller hears from them again. While on that list, it's possible the hub won't even try to send messages to that device (this part I am not sure of, but presumably it might try to limit things it know won't work to keep the network happy, like it appears OpenZWave does, the only implementation whose behavior I'm really aware of). Again, this is just a guess on my part--I'd love to hear what anyone might know.
Also, it seems there are two issues being conflated above: "ghost nodes" (nodes that are on the controller but don't have an associated device in Hubitat) and ... whatever we want to call nodes that do exist in both places but are out-of-service as far as the real-world is concerned. I've seen a few users, including some above, report that the former has caused problems for them, possibly hub problems (slowdowns? not sure). While the latter may definitely cause network problems (e.g., a device trying to route through a repeater that no longer really exists but everything else still thinks is there), I'm not sure if I've seen anyone mention hub problems in that regard. For that matter, I'm not really sure that the former are hub problems, as network problems might have overlapping symptoms (slowness) if all your devices are Z-Wave.
Hubitat doesn't currently provide a way to see "ghost nodes" in the former sense, but third-party software like Z-Wave PC Controller with a Z-Wave stick (either one you're already using as the hub's primary controller or one you join as a secondary) can. There isn't a good way to see "missing" nodes in the latter sense, aside from clues that device activity may give you, but those problems are usually easier to figure out: if you added a device and then unplugged it, then you already know. (And again, I'm not denying that this could cause problems, and it's something I avoid if I can...just pointing out that some people are talking about two different things.)