adding more options to control the size of the GlobalTriggerDecision.
Currently, the bottleneck seems to be the streaming of the output object,
it is much slower if many input objects have been added. New options:
-forward-input forwards the input object instead of adding them
-include-input=none do not include anything in the decisionobject
-include-input=short only include a short TNames array which describes the
objects
-include-input=objects include all the objects, this is the current behavior
and remains the default