Add xxhash64 support for nested types

[ustcfy]

Contribute to #2574

This PR adds support for xxhash64 with struct , list and map types.

The hardcoded expected values in the tests are generated using Spark.
For example, you can produce the same results as in testXXHash64ListOfStruct using the following code.

import org.apache.spark.sql.functions._
import org.apache.spark.sql.types._
import org.apache.spark.sql.Row

val schema = StructType(Seq(
    StructField("listofstruct",
        ArrayType(
            StructType(Seq(
                StructField("string", StringType, true),
                StructField("int", IntegerType, true),
                StructField("double", DoubleType, true),
                StructField("float", FloatType, true),
                StructField("bool", BooleanType, true)
            )),
            true
        )
    )
))

val data = Seq(
    Row(List()),
    Row(List(Row("a", 0, 0.0, 0f, true))),
    Row(List(Row("B\n", 100, 100.0, 100f, false), Row("dE\"\u0100\t\u0101 \ud720\ud721", -100, -100.0, -100f, null))),
    Row(List(Row("A very long (greater than 128 bytes/char string) to test a multi hash-step data point " +
        "in the MD5 hash function. This string needed to be longer.", Int.MinValue, java.lang.Double.longBitsToDouble(0x7ff0000000000001L), java.lang.Float.intBitsToFloat(0xff800001), false))),
    Row(List(Row(null, Int.MaxValue, java.lang.Double.longBitsToDouble(0x7fffffffffffffffL), java.lang.Float.intBitsToFloat(0xffffffff), true), Row(null, null, null, null, null))),
    Row(null)
)

val df = spark.createDataFrame(spark.sparkContext.parallelize(data), schema)
df.columns.foreach(c => println(s"$c => ${df.select(xxhash64(col(c))).collect.mkString(",")}"))

You will get the following output:

listofstruct => [42],[7451748878409563026],[948372773124634350],[8444697026100086329],[-5888679192448042852],[42]

[res-life]

Please add more cases, LGTM for the function.

[res-life]

LGTM

[res-life]

build

[revans2]

looks good, but I am no expert on the CUDA side of things.

[revans2]

Can we have a follow on issue to make the hash function a template? We have the same issues with all hash functions we want to support, so having generic code would be nice.

[firestarman]

Look good to me, but a nit.

[ttnghia]

Sorry, this PR has similar implementation with the HiveHash one, so both need to address the issue with shallow stack size unless we want to target this for 24.12.

The base branch was changed.