feat: support `cast StructType to String` when duplicated column in the struct

[comphead]

Describe the bug

The cast StructType to StringType fails on duplicated columns with auto and native_scan

          checkSparkAnswerAndOperator(
            "SELECT CAST(struct(_12, _12) as string) FROM tbl")

Expected only Comet native operators, but found Project.
plan: *(1) Project [cast(struct(_12, _12#11, _12, _12#11) as string) AS CAST(struct(_12, _12) AS STRING)#59]
+- *(1) CometColumnarToRow
   +- CometScan [native_iceberg_compat] parquet [_12#11] Batched: true, DataFilters: [], Format: CometParquet, Location: InMemoryFileIndex(1 paths)[file:/private/var/folders/6f/_s1vnnd55zgfkx7zlwnrnv0h0000gn/T/spark-76..., PartitionFilters: [], PushedFilters: [], ReadSchema: struct<_12:decimal(20,0)>

Found when working on #2175

Steps to reproduce

No response

Expected behavior

No response

Additional context

No response

[comphead]

This happens because of

        if (struct.names.length != struct.names.distinct.length) {
          withInfo(expr, "CreateNamedStruct with duplicate field names are not supported")
          return None
        }

Having commented this out I'm getting

Caused by: java.lang.IllegalStateException: ArrowArray struct has 2 children (expected 1)

[andygrove]

@comphead It looks like the current behavior is correct, because we fall back to Spark when there are duplicate fields? If so, can we remove the bug label?

[andygrove]

Closing this as not planeed for now. We have correct fallback behavior. Let's re-open if someone really needs this for performance reasons.