as can be derived from the file names, those SQL scripts are generated code.
After schema changes (e.g. added columns, but also added tables) the SCI views have to be (re)created.
It's easy to detect changes, but it's a bit of a hassle to distinguish between a table change and a new table.
Hence if a change is detected, a DROP VIEW and CREATE VIEW is generated, well knowing that the DROP VIEW will fail in case of a new table.
I'd love to have a kind of "EXISTING" keyword here.
But in the end it doesn't make a difference: we want to get rid of the original view and if it doesn't exist (any more) after executing the DROP statement, we're fine.
The important statement, the CREATE VIEW, will be successful afterwards.
Given the fact that those scripts are executed only once, we accept the ugly output but correct behaviour.