Explore if GDM censoring can be simplified

[Aastedet]

If performance is similar with only end-of-pregnancy dates for censoring GDM, it would allow us to remove the maternal care visits to make the algorithm and data requirements simpler.

It would also streamline the algorithm if/when we eventually add support for using Medical Birth Register data to censor GMD instead of diagnoses from the National Patient Register, as both data sources would ultimately just provide one variable with date of the end of each pregnancy for each ID.