[SPARK-38775][ML] cleanup validation functions

[zhengruifeng]

What changes were proposed in this pull request?

1, remove unused extractInstances and extractLabeledPoints in Predictor;
2, remove unused checkNonNegativeWeight in function;
3, move getNumClasses from Clasifier to DatasetUtils;
4, move getNumFeatures from MetadataUtils to DatasetUtils;

Why are the changes needed?

to unify to methods

Does this PR introduce any user-facing change?

No

How was this patch tested?

existing testsuites

[srowen]

This is just refactoring - shouldn't change behavior? seems OK

[srowen]

getOrElse works here too but doesn't matter

[zhengruifeng]

ok, will swith back to getOrElse

[dongjoon-hyun]

Hi, @zhengruifeng and @srowen (also cc @gengliangwang )

If you don't mind, could you hold on this refactoring a little bit more until @MaxGekk releases RC1?

1. Currently, Apache Spark 3.3 is under active testing and unstable.2.
2. GitHub Action PR builder also doesn't provide you the ANSI test result
3. This kind of massive refactoring increases the complexity during the Apache Spark 3.3.0 release process.
   • For example, SPARK-38490 (#35797) introduced a ANSI conf related bug and SPARK-38776 fixed it. However, SPARK-38669 (#35983) refactored master and branch-3.3 different a lots and the backporting was difficult.

Hopefully, please give us more time until we finish Apache Spark 3.3, @zhengruifeng . I believe Cleanups and refactoring PRs are less urgent than the scheduled release.

[zhengruifeng]

@dongjoon-hyun Ok, I will hold on this PR since its target version is 3.4

[dongjoon-hyun]

Thank you so much!

[srowen]

We can revisit this now. Rerun tests I think?

[dongjoon-hyun]

+1 for restarting. Thank you, @zhengruifeng and @srowen .

[zhengruifeng]

Sure, let me update this PR

[dongjoon-hyun]

+1, LGTM. Thank you, @zhengruifeng , @srowen , @huaxingao .
Merged to master.

[zhengruifeng]

Thanks all!