cache: improve service discovery reliability

cache: improve service discovery reliability

Send queries for non-host PTR services out to all interfaces instead of
just the one that they were first discovered on.
When updating, go through all non-host service entries and send a query.
When host records expire while non-host service entries remain, queries
were not automatically re-sent, causing some services with lower TTL to
be missed.

Signed-off-by: Felix Fietkau <nbd@nbd.name>