Update spark dependency to spark 3.2

[WeichenXu123]

Current xgboost4j released version built with spark 3.0, but the binary jar is incompatible with spark 3.2
This PR update the spark dependency to be spark 3.2

[WeichenXu123]

@CodingCat Could you take a look ? Thanks!

[codecov-commenter]

Codecov Report

Merging #7359 (9ac66c0) into master (a6bcd54) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #7359   +/-   ##
=======================================
  Coverage   83.71%   83.71%           
=======================================
  Files          13       13           
  Lines        3892     3892           
=======================================
  Hits         3258     3258           
  Misses        634      634           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a6bcd54...9ac66c0. Read the comment docs.

[WeichenXu123]

Gentle ping @CodingCat , Thanks!

[trivialfis]

@wbo4958 Could you please help take a look?

[wbo4958]

Ok, I will verify it. Thx @WeichenXu123

[wbo4958]

I just verified it with xgboost1.5.0 release, which will throw exception running on spark 3.2.0 release.

The exception happened in saving a model

Exception in thread "main" java.lang.NoSuchMethodError: org.json4s.JsonDSL$.pair2Assoc(Lscala/Tuple2;Lscala/Function1;)Lorg/json4s/JsonDSL$JsonAssoc;
	at ml.dmlc.xgboost4j.scala.spark.params.DefaultXGBoostParamsWriter$.getMetadataToSave(DefaultXGBoostParamsWriter.scala:75)
	at ml.dmlc.xgboost4j.scala.spark.params.DefaultXGBoostParamsWriter$.saveMetadata(DefaultXGBoostParamsWriter.scala:51)
	at ml.dmlc.xgboost4j.scala.spark.XGBoostClassificationModel$XGBoostClassificationModelWriter.saveImpl(XGBoostClassifier.scala:522)
	at org.apache.spark.ml.util.MLWriter.save(ReadWrite.scala:168)
	at org.apache.spark.ml.util.MLWritable.save(ReadWrite.scala:287)
	at org.apache.spark.ml.util.MLWritable.save$(ReadWrite.scala:287)
	at ml.dmlc.xgboost4j.scala.spark.XGBoostClassificationModel.save(XGBoostClassifier.scala:214)
	at iris$.main(iris.scala:53)
	at iris.main(iris.scala)

But when I updated the spark dependency to 3.2.0 and then built xgboost, the xgboost can run well on spark 3.2.0 but it will fail on spark versioned before 3.2.0

So we need a way to be compatible all 3.0+ spark.

[trivialfis]

Is that a binary compatibility issue?

[wbo4958]

Hi @WeichenXu123, I submitted another PR for the json4s compatible issue. Please help to check it

#7376

[wbo4958]

@trivialfis, looks like we can close this PR

[trivialfis]

Thank you both!