The logic that generates IDs for DMN evaluations should be pluggable via custom beans.

This allows for more complex scenarios like advanced tracing of DMN models being executed in processes via correlation IDs (which are out of the scope of this ticket).