We are considering moving to Delta Live tables from a traditional sql-based data warehouse.

Worrying me is this FAQ on identity columns Delta Live Tables frequently asked questions | Databricks on AWS this seems to suggest that we basically can't create unique ids for rows unless streaming and of course a SCD 1 dimension gold table seems like it will never be able to be a streaming table as it is not an append-only (it needs to merge)

Databricks talks up the requirement here .. How to Identity Columns to Generate Surrogate Keys in the Databricks Lakehouse - The Databricks Blog so it is clear how important this is. 

What am I missing here?