Refactoring DnsCache and how to resolve Service Instance (#108)

Simplify the logic of resolving an instance: always resolve from the cache.
For any incoming updates, we update the cache first, then try to resolve.

src/dns_parser.rs

@@ -73,6 +73,9 @@ pub(crate) struct DnsRecord {
     pub(crate) entry: DnsEntry,
     ttl: u32,     // in seconds, 0 means this record should not be cached
     created: u64, // UNIX time in millis
+
+    /// Support re-query an instance before its PTR record expires.
+    /// See https://datatracker.ietf.org/doc/html/rfc6762#section-5.2
     refresh: u64, // UNIX time in millis
 }
 
@@ -88,6 +91,10 @@ impl DnsRecord {
         }
     }
 
+    pub(crate) fn get_created(&self) -> u64 {
+        self.created
+    }
+
     pub(crate) fn is_expired(&self, now: u64) -> bool {
         get_expiration_time(self.created, self.ttl, 100) <= now
     }
@@ -796,6 +803,8 @@ pub(crate) struct DnsIncoming {
     offset: usize,
     data: Vec<u8>,
     pub(crate) questions: Vec<DnsQuestion>,
+    /// This field includes records in the `answers` section
+    /// and in the `additionals` section.
     pub(crate) answers: Vec<DnsRecordBox>,
     pub(crate) id: u16,
     flags: u16,