audit: Try to send audit even if the status is offline (#18458)

Currently, once the audit becomes offline, there is no code that tries
to reconnect to the audit, at the same time Send() quickly returns with
an error without really trying to send a message the audit endpoint; so
the audit endpoint will never be online again.

Fixing this behavior; the current downside is that we miss printing some
logs when the audit becomes offline; however this information is
available in prometheus

Later, we can refactor internal/logger so the http endpoint can send errors to
console target.

internal/logger/target/http/http.go

@@ -379,10 +379,6 @@ func (h *Target) SendFromStore(key store.Key) (err error) {
 // Messages are queued in the disk if the store is enabled
 // If Cancel has been called the message is ignored.
 func (h *Target) Send(ctx context.Context, entry interface{}) error {
-	if atomic.LoadInt32(&h.status) == statusOffline {
-		h.config.LogOnce(ctx, fmt.Errorf("target %s is offline", h.Endpoint()), h.Endpoint())
-		return nil
-	}
 	if atomic.LoadInt32(&h.status) == statusClosed {
 		return nil
 	}