This PR addresses #12943

Every other transform has 2 public overloads: one that takes a targetName, and one that does not. The one exception is identity, where the overload with targetName is not public. This looks like an oversight. This functionality is needed to support partition transforms with renamed source columns correctly.

Concretely, here is a use case where this is necessary:

• Engine A: CREATE TABLE t (col1 int) PARTITION BY (col1) -> creates a transform called col1
• Engine A: ALTER TABLE t RENAME col1 TO col2 -> transform is still called col1
• Engine B: Tries to read or write t. At this point Engine B must be able to create a transform called col1 which has col2 as the source in order to accurately reflect Iceberg metadata.