fix: use functionForEval to unify lambda variable references

The raw LambdaFunction body may contain a different NamedLambdaVariable
instance than the one extracted from arguments. Using functionForEval
(as Spark does internally) ensures the variable we set is the same
instance the body reads from.

Author: andygrove

common/src/main/scala/org/apache/comet/udf/ArrayExistsUDF.scala

@@ -56,7 +56,8 @@ class ArrayExistsUDF extends CometUDF {
     val registryKey = new String(keyVec.get(0), StandardCharsets.UTF_8)
     val arrayExistsExpr = CometLambdaRegistry.get(registryKey).asInstanceOf[ArrayExists]
 
-    val LambdaFunction(body, Seq(elementVar: NamedLambdaVariable), _) = arrayExistsExpr.function
+    val LambdaFunction(_, Seq(elementVar: NamedLambdaVariable), _) = arrayExistsExpr.function
+    val body = arrayExistsExpr.functionForEval
     val followThreeValuedLogic = arrayExistsExpr.followThreeValuedLogic
     val elementType = elementVar.dataType