Update European Parliament corpus

[lukavdplas]

Update to the EU parliamentary corpus. I've indexed the new data so this is ready to be released.

This update uses a different dataset for the older data (pre-2024). Compared to the current dataset (Talk of Europe), this dataset (EUPDCorp) covers a longer period, is easier to parse, and includes translations consistently.

For the post-2024 data, which is extracted from the EU's Open Data API, there are several fixes and improvements:

• Speech text is now the full text, instead of just the first paragraph.
• Greatly reduces the number of requests to the API which speeds up processing, as requests can take a long time. (The full job now takes hours instead of days.) The reader no longer cold-calls debate IDs to check if there was a debate on that day, and requests speeches in batches of 50 instead of one at a time.
• Also extracts the original language text.
• Extract more speaker metadata, including birth year, gender, and national party.
• Various fixes for missing keys, null values, and other things that trip up extraction.

This branch includes some features that should be moved to ianalyzer_readers, namely:

• An RDSReader (with additional Python dependency)
• Change to the JSON extractor.
  I've opened PRs for that (RDS reader ianalyzer-readers#42, allow empty list for JSON keys ianalyzer-readers#43) and will remove the code here when it is released.

Issues:

• Resolves New European Parliament dataset #1802
• Resolves European Parliament: incorrect links #1951 (which is no longer applicable)