Questions I am trying to figure out Why does the optimizer work better (not perfect) when flattenAssociativeOperations is applied to the AST before optimization?