Have you tried something like this already? 

 Force Cache Invalidation (Recommended)

  -- After CREATE OR REPLACE VIEW, execute:

  ALTER SHARE <share_name> REMOVE TABLE <schema>.<view_name>;

  ALTER SHARE <share_name> ADD TABLE <schema>.<view_name>;

  This forces DSFF to drop materialized views and recreate them on next access.

You could also automate this with SDK. 

# Use Databricks SDK to trigger cache refresh

  from databricks.sdk import WorkspaceClient

  def invalidate_dsff_cache(share_name, view_name):

      w = WorkspaceClient()

      # Remove and re-add to share

      w.shares.update(

          name=share_name,

          updates=[

              RemoveTable(name=f"{schema}.{view_name}"),

              AddTable(name=f"{schema}.{view_name}")

          ]

      )