They finally kicked me over to another department (tech guys I think) who found that a previous tenant, years earlier, had the emergency only (life-line) service. It had been "disconnected" in the system in every way as far as billing and such were concerned, but wasn't actually physically disconnected. The tech guys were finally able to fix it. (...) This is a case where you'd think their system would be able to detect that calls were being placed by a residence that had no service. Nope.
I don't know how it is in the US, but here in Norway you can dial our equivalent of 911 from any cell phone, connecting to any tower in range regardless if it has service or even a SIM card and I assume landlines work on the same principle. That the service was "disconnected" just means they don't have any obligation to keep it working, but they're not going to block any 911 call ever, I don't know if there's a law but the bad publicity would be a disaster. So this is probably by design and a feature, not a bug but customer service was probably not aware of this.
They might not even have access to the raw call log, since their user interface probably revolves around services and calls tied to a service. After all how often do you have a phone line with no service dialing 911 - because that's the only place you can reach - then calling customer service complaining that the call came through? This was a 0.01% corner case and I'm not surprised they thought it "impossible" and had to escalate to someone with real knowledge of the inner workings of the physical network.