Set loopback="true" in all JGroups UDP stacks

FLUSH and MuxChannel service view messages need to be received by the sender. If there is a problem with multicast on the interface the channel is using, the messages will not be received, leading to long delays and incorrect behavior.

Per Bela, it is fine (desirable) to set loopback="true" in the transport protocol in each stack. This will ensure the channel receives its own messages.

If multicast isn't working, the channel will of course not be able to communicate with other members, and thus isn't really working properly. But, debugging and resolving these sorts of connectivity issues is something well documented in JGroups docs, wikis etc. The new weird behavior you get when FLUSH and MuxChannel can't receive their own messages is not easy to understand.