Validation of documents

[lfoppiano]

As emerged in task #73 when a document is being harvested and it has incompleted fields (e.g. no title) it needs to be validated at some point in the pipeline.
The validation could result in the document being accepted, discarded, queued on a separate storage where domain expert decide how to threat it (discard it, complete the data, do nothing, etc.).

Here example of document without title: https://hal.archives-ouvertes.fr/hal-01337963v1/tei

Ideally, a separate module should take care of this validation.