OPTIM: resolvers: move the eb32 node before the data in the answer_item

perf top shows that we spend a lot of time trying to read item->type in
the lookup loop, because the node is placed after the very long name,
so when the node is found, no data is in the cache yet. Let's simply
move the node upper in the struct. This results in the CPU usage of
resolv_validate_dns_response() to drop by 4 points.
diff --git a/include/haproxy/resolvers-t.h b/include/haproxy/resolvers-t.h
index 48bff0c..2e53d91 100644
--- a/include/haproxy/resolvers-t.h
+++ b/include/haproxy/resolvers-t.h
@@ -111,6 +111,7 @@
 	uint16_t        weight;                      /* SRV type weight */
 	uint16_t        port;                        /* SRV type port */
 	uint16_t        data_len;                    /* number of bytes in the <data> field below */
+	struct eb32_node link;                       /* linking node */
 	union {
 		struct sockaddr_in in4;              /* IPv4 address for RTYPE_A */
 		struct sockaddr_in6 in6;             /* IPv6 address for RTYPE_AAAA */
@@ -119,7 +120,6 @@
 	unsigned int    last_seen;                   /* When was the answer was last seen */
 	struct resolv_answer_item *ar_item;          /* pointer to a RRset from the additional section, if exists */
 	struct list	attached_servers;            /* attached server head */
-	struct eb32_node link;                       /* linking node */
 };
 
 struct resolv_response {