Update Doc2Vec models

[jmsevin]

In order to include Covid-19 concepts in results and test document embeddings in higher dimension:

• [x] Download recent Wikipedia corpora
• [x] Train Doc2Vec model for English in 500 dimensions
• [x] Train Doc2Vec model for French in 500 dimensions
• [x] Train Doc2Vec model for Spanish in 500 dimensions
• [x] Compute new similarity score thresholds
• [x] Upload new models on production server

[jmsevin]

As new models need more space, this issue needs the storage issue on production server to be solved.