interface scanning happens much too often

The route socket is supposed to detect changes in configured interfaces by listening for RTM_NEWADDR and RTM_DELADDR messages; these messages trigger a call to ns_interface_scan() which causes the listening interfaces to be updated.

On my system, such messages are sent every few seconds, and there seems to be one for each configured address on the system. So in the log I see bursts that look like this:

14-Dec-2021 00:05:02.196 route_recv: success
14-Dec-2021 00:05:02.200 route_recv: success
14-Dec-2021 00:05:02.200 route_recv: success
14-Dec-2021 00:05:02.200 route_recv: success
14-Dec-2021 00:05:02.200 route_recv: success
14-Dec-2021 00:05:02.200 route_recv: success
14-Dec-2021 00:05:02.204 route_recv: success
14-Dec-2021 00:05:02.204 route_recv: success
14-Dec-2021 00:05:02.204 route_recv: success
14-Dec-2021 00:05:02.204 route_recv: success
14-Dec-2021 00:05:02.308 route_recv: success
14-Dec-2021 00:05:02.308 route_recv: success
14-Dec-2021 00:05:02.312 route_recv: success
14-Dec-2021 00:05:02.312 route_recv: success
14-Dec-2021 00:05:02.312 route_recv: success
14-Dec-2021 00:05:02.316 route_recv: success
14-Dec-2021 00:05:02.316 route_recv: success
14-Dec-2021 00:05:02.316 route_recv: success
14-Dec-2021 00:05:02.316 route_recv: success
14-Dec-2021 00:05:02.320 route_recv: success

... and each of those causes the interfaces to be re-scanned. And then it happens again 3-5 seconds later.

This wasn't a fatal error until #3053 (closed) was merged, which caused re-scanning of interfaces to be destructive - each of those events will cause TLS and HTTPS interfaces to be shut down and recreated. But that's a separate bug I'll report it as a separate issue. (#3056 (closed))