fix: downgrade fluentbit to fix tls.vhost issues (#4871)

When testing with TLS, we found the `tls.vhost` configuration for
Fluent Bit HTTP outputs does not work as it did before, causing Fluent
Bit to fail with TLS errors.

(cherry picked from commit 2f0464f)

master/pkg/aproto/defaults.go

@@ -1,6 +1,8 @@
 package aproto
 
 const (
-	// FluentImage is docker image id to use for the managed Fluent Bit daemon.
-	FluentImage = "fluent/fluent-bit:1.9.3"
+	// FluentImage is docker image id to use for the managed Fluent Bit daemon. Fluent Bit did a
+	// rewrite of their TLS stack that introduced a bug (unconfirmed with them) with `tls.vhost`
+	// in 1.7, so take care before upgrading this past that (1.9.3 did not work).
+	FluentImage = "fluent/fluent-bit:1.6"
 )