[BUG] ORC read can corrupt data when specified schema does not match file schema ordering

[wbo4958]

@firestarman has found the same issue with #3007 for the special schema which can't be pruned. I can repro it with the below code.

      val df = Seq(Testing(1, "hello", 2021)).toDF
      df.printSchema()
      // root
      // |-- _col1: integer (nullable = false)
      // |-- _col2: string (nullable = true)
      // |-- _col3: long (nullable = false)
      df.show()
      // +-----+-----+-----+
      //|_col1|_col2|_col3|
      //+-----+-----+-----+
      //|    1|hello| 2021|
      //+-----+-----+-----+
      df.write.mode("overwrite").orc(resource1)

      val schema = StructType(
        Seq(
          StructField("_col2", StringType),
          StructField("_col3", LongType),
          StructField("_col1", IntegerType),
          ))
      val dfRead = spark.read.schema(schema).orc(resource1)
      dfRead.show()

The GPU output is

+-----+-----+-----+
|_col2|_col3|_col1|
+-----+-----+-----+
|     |    1|    5|
+-----+-----+-----+

while the CPU output is

+-----+-----+-----+
|_col2|_col3|_col1|
+-----+-----+-----+
|    1| null| 2021|
+-----+-----+-----+

Looks like there also is an issue for CPU reading ORC