Well, I’ll reply to my own post. Maybe this will help someone in the future.
I managed to get the device “configured.” I’m still not entirely clear exactly what that means. I’m a twenty+ year software developer, and I have a strong need to understand this at a level of detail that is apparently not documented anywhere, which is unfortunate. All I can say for sure is that unless the device is “configured,” it will show up with a red circle in the upper left corner of its dialog, and one of several error messages below it (“Device unable to be configured” or “Unable to get any information on node”). Those error messages do not appear to affect the device’s operation, however. It responded to on/off messages just fine.
To get my device (a Leviton VRCS2) configured, I first excluded it from the Z-wave network, then did a factory reset on it, then re-included it. For this step, I brought the Vera to within a few inches of the switch. My laptop was directly connected to the Vera, and I did not use the battery mode, but I’m not sure if that was critical. I wanted to remove all possible variables.
After they were included on the network, I changed their names and the rooms and saved that. Then I configured them, going to wrench → Settings → Configure this node now. I did this, again, with the Vera just a few inches away from the switch. It’s a 2-switch device, so I had to do each one in turn. Note that much of the documentation here (which is the ONLY documentation) said that I would see a scene controller and 2 switchable loads. This scene controller never showed up while pairing. But this may be a blessing, since everything I’ve seen says that once you do get this device configured, you then have to go through the trouble of creating 4 scenes if you still want to turn the lights on and off via the switch buttons in the usual way. So, now, I have two switches for the two loads that work either manually or remotely in the intuitive way.
It took a couple of tries to get them configured, but eventually it worked. And then I noticed something: a few minutes later, for a reason I don’t understand and expect to never know, they started trying to reconfigure themselves. And failing, because now, the Vera was maybe 3 feet away. I held it back up in close proximity, got them configured, and then changed them to stop automatically configuring on the Settings page. This seems to have worked. The only problem I can see is that the timestamp of when they were last configured is from 31-Dec-1969 because the Vera was not on the internet when I got them configured.
So, right now, I have what I want: a dual switch controlling two sets of overhead lights in a room, where the buttons work as I’d expect and I can control them remotely.
I also have a lot of frustration, which I should maybe save for another time. But in general, I think it’s unfair to market the Vera as a consumer device for the general population. The quality of the software and the scarcity of the documentation make this feel much more like a homebrew/DIY type system. I think a very, very tiny sliver of the population wants to deal with ssh’ing their way into the box and tailing something in /var/log/cmh to understand what’s going on. I’m going to be looking into alternative controllers; this feels too much like my day job.