There's a few things this piece of code is trying to do:

1. If we have an aggregate exception, try to update the bucket
   parameters to represent the stuff inside. Rather than doing that
   it's much easier to just do a report for each inner exception, which
   will not interfere with other targeted notification policies.
2. If we have a remote exception, try to bucket on some other
   information. What I don't get is the information there is the stack
   trace (which is the client stack trace) and the message which we'd
   already include anyways -- nothing there is any more actionable
   than just falling into the default error reporting logic.
3. Trying to report inner exceptions over the outer exception if
   ever one exists. I can't say that's a great idea unless we actually
   know the type of exception more directly.

Just deleting this all seems much easier.