Bulk Service Processor should be contributed by configuration

Today the Bulk Service Processor (the scroller and status computations) is created by code because there are dynamic parts (the action source stream) and the initialization must be controlled:
1. Bulk Service streams must be created: command, status and done streams
2. Action processors source streams must be created: for instance csvExport ...
3. Bulk Service computation must be started
4. Action processors can be started

Having a way to register a processor and initialize its streams and be able to mark streams as being external (defined in another processor) will allow to create the Bulk Service processor like any other processor using configuration.