Add heuristics for finding incorrect dates in the protocols.

[BobBorges]

At the library, they want to have exact dates for all the protocols. We know there are some incorrect dates in the protocols due to:

• cover sheets where the date for several protocols are written
• running headers of protocol compilations
• protocols that start/end on the same physical page
• accidentally scraping references to dates that aren't a document date

We need to work on eliminating them by adding heuristic tests to the workflow

• Dates shouldn't overlap from one protocol to the next
• until 1874 date is part of filename
• date scraping should ignore dates associated with sjükanmalan läkarintyg / läkarbytyg / sjükbetyg + signature of the läkare who signed off on the sickness
  • code to manage this is here and here
• date spans are < 1 week

What else?

[BobBorges]

sjuk betyg example

[BobBorges]

• identify start and end of protocols
• kick leading and trailing text

cf.:

• Decision 6: Single motion content per file the-swedish-parliament-corpus#26
• Remove duplicate text in pre-1872 protocols welfare-state-analytics/riksdagen-corpus#327