Ticket #218 - Make RI Plugin worked with replicated updates

Bug Description:  If a user is deleted from AD (or a master not using RI Plugin),
                  the operation comes in as a replicated op, and RI aborts its
                  processing.

Fix Description:  Added a new config attribute to the plugin "nsslapd-pluginAllowReplUpdates: on".
                  We check for this setting at startup and set a flag.  We then allow these
                  operations, and we also attach OP_FLAG_REPLICATED flag to pblock for
                  slapi_modify_internal().

Side note:        Cleaned up all the indentation/comments in referint.c

https://fedorahosted.org/389/ticket/218

reviewed by:

ldap/servers/plugins/replication/repl5_plugins.c

@@ -479,7 +479,12 @@ multimaster_preop_modify (Slapi_PBlock *pb)
             }
             else
             {
-                PR_ASSERT(0); /* JCMREPL - A Replicated Operation with no Repl Baggage control... What does that mean? */
+                /*  PR_ASSERT(0); JCMREPL - A Replicated Operation with no Repl Baggage control... What does that mean? */
+                /*
+                 *  This could be RI plugin responding to a replicated update from AD or some other master that is not
+                 *  using the RI plugin, so don't PR_ASSERT here.  This only happens if we configure the RI plugin with
+                 *  "nsslapd-pluginAllowReplUpdates: on", also the RI plugin only issues "modifies".
+                 */
             }
         }
         else