Test Case for Subscription Event Bug

[swalker2m]

There's a report that a simple update to S/N requirements generates up to 13 observation edit events. This PR will attempt to find the issue.

[swalker2m]

This may not be a great idea after all. By the time the event is fired, the expected subscription query can produce different results than it would have if it had been generated immediately. For example, when a new group is created without specifying a parent index, it defaults to index 0. It may be subsequently moved. Then the group creation event fires after the GroupingTimeout expires. At that point, the index is no longer 0. So you can see a CREATED event with parent index 1 followed by an UPDATED event also at 1 when you expected CREATED at 0 followed up UPDATED showing 1.

{
  "groupEdit" : {
    "editType" : "CREATED",
    "value" : {
      "id" : "g-10c",
      "parentId" : "g-10b",
      "parentIndex" : 1
    }
  }
},
{
  "groupEdit" : {
    "editType" : "UPDATED",
    "value" : {
      "id" : "g-10c",
      "parentId" : "g-10b",
      "parentIndex" : 1
    }
  }
}

I don't know whether this is a problem for production, but it definitely makes tests unpredictable.