Configuring a remote EJB client, whether standalone or within AS, to use
RetryInterceptor (RI) when EJBs are deployed across different node/clusters
can be a bit tricky cos jndi details to be used by interceptor are stored statically.

There're several available workarounds which I'll explain in the workaround
section. This JIRA is focused at the feasibility of creating a more intelligent RI
that caches JNDI lookups as per EJB type.