Add flatten processor

[MKLeb]

I'm thinking it may be a good idea to add a processor to flatten the data dictionary... This could be helpful in producing expected formats between steps, and especially before the forwarding stage of pipelines.