Support multiple datasets from Arkindex as input

Closes #244 (closed)

changed milestone to %DAN-P6: Training improvements Part 2

added P2 label

assigned to @mblanco

requested review from @yschneider

This works for extracting multiple datasets in a single corpus/database. We'll also need to support multiple databases. We'll talk later to see how we could do that efficiently.

resolved all threads

mentioned in issue #239 (closed)

merged

mentioned in commit 4cf5e694