[Spark] Handle NullType in normalizeColumnNames()

The sanity check in normalizeColumnNamesInDataType() introduced by that change is a bit too restrictive, and fails to handle NullType correctly.

Closes #2634

GitOrigin-RevId: faaf3d981c57ef3ceb4081e0bc94d457359fc9d8

spark/src/main/scala/org/apache/spark/sql/delta/schema/SchemaUtils.scala

@@ -256,6 +256,10 @@ def normalizeColumnNamesInDataType(
           keyType = normalizedKeyType,
           valueType = normalizedValueType
         )
+      case (_: NullType, _) =>
+        // When schema evolution adds a new column during MERGE, it can be represented with
+        // a NullType in the schema of the data written by the MERGE.
+        sourceDataType
       case _ =>
         if (Utils.isTesting) {
           assert(sourceDataType == tableDataType,