AlterTypeStatement generates potentially huge number of schema announcements, only one of them necessary

Arguably an oversight from my 3.0 schema work. In 3.0 and up we store type names - including UDTs - by their name only in all schema tables, including columns, functions, and other types.

As such, it's completely redundant to issue a migration for each affected table/function/type, the original type change migration will take care of everything automatically. It's dead and harmful code.