Since DNPM Datenmodell 2.1 changes from time to time,
use a (mostly) autogenerated library based on JSON schema.

This is required if async messaging is used. The request is already gone if response gets processed.

Workflow:

* Keep old followup_count
* Add expected_followup_count
* If: DNPM:DIP accepts the request - copy to to followup_count
* Else: keep old followup_count

This prevents from saving unaccepted follow-up counts.

This is a first approach to make use of a reason missing given in MTB file at UKGM Marburg due to special handling of research consent.

Without these changes, the request will be canceled without any further processing.

This is done in addition to the check by date.
The submission type will be set to "followup" if:

* It is not an initial submission
* There are follow-ups in submission content
* The date or the count is less than the last successful submission

In the database, existing requests will be initialized with
followup_count = -1.

This identifies follow-up submissions:

* if the initial submission has been accepted
* if the most recent follow-up is later than the patients last submission

If these criteria are not met, the submission will be sent as an addition