SpecificEntities with DeepNetwork models for other WikiNER languages

The Deep pipeline for NamedEntity Recognition in Fre and Eng is based on NLP models trained on the [WikiNER datasets](https://figshare.com/articles/Learning_multilingual_named_entity_recognition_from_Wikipedia/5462500).

Since DeepLima povides tokenisation for more languages, we should use the other languages of WikiNER] to train new models :

- aij-wikiner-en . done
- aij-wikiner-fr . done
- aij-wikiner-de . todo
- aij-wikiner-es . todo
- aij-wikiner-it . todo
- aij-wikiner-nl . todo
- aij-wikiner-pl . todo
- aij-wikiner-pt . todo
- aij-wikiner-ru . todo

Start by locating the Deep pipeline's named-entity-recognition model-training and model-integration entry points, then review the linked WikiNER dataset and the completed English and French models. Confirm how the German, Spanish, Italian, Dutch, Polish, Portuguese, and Russian models should be trained and registered; done means those listed todo languages have usable integrated models.

Written by the indexing model from the issue text.