Rewrite intake stage from WDL to Python Polars

[vincent-octo]

TODO

• Encapsulate the intake stages into a mini WDL script?
  Out of scope for this PR.
• Documentation on how to run the intake.assemble and intake.tidyup stages
• Maybe provide a single command to run both intake stages at once
• Add a sanitization step to remove or replace new-line characters, as users are used to not having to parse new-line characters. Maybe a good test for that is, can the output be read with non-CSV-aware tools like basic awk.
• Compare new output (this rewrite) to previous output, expectation is that they match, document the differences otherwise.
  • N rows
  • columns names and order
    Diff:
    • ROW_ID is now _rowid
    • new _rowid_source column to have end-to-end data tracing
  • N rows for each FINNGENID
  • Exact value match for maximal selection of columns (= all except row ID columns)
    Diff:
    • Behavior change: Newline inside values are now replaced by the Unicode character ␤ (U+2424), instead of replacing by a space character .
    • Bug fix: Values within quoted TSV fields are now correctly preserved, whereas the previous implementation added extra quotes.
    • Bug fix: Tab characters inside quoted TSV values are now correctly preserved, whereas the previous implementation treated them as field separator resulting in shifted values.