[SPARK-28958][ML][PYSPARK][WIP] pyspark.ml function parity

[zhengruifeng]

What changes were proposed in this pull request?

1, remove setters generated by _shared_params_code_gen.py;
2, add common abstract classes like the side side, such as JavaPredictor/JavaClassificationModel/JavaProbabilisticClassifier;
3, for each alg, add its param trait, such as LinearSVCParams;
4, since sharedParam do not have setters, we need to add them in right places;

Why are the changes needed?

yes, some methods in estimators and models in the py side are changed

How was this patch tested?

I test them in pyspark repl for now

[zhengruifeng]

Since I remove the setters generated by _shared_params_code_gen.py, it seems that all algorithms(classification/regression/clustering/als/features/fpm/etc) needed to be updated in one batch.
This PR is now incomplete, and only modify classification&regression for now, because in this way I need to change too many place in a single PR. I wander if there is a better way to do this.

[zhengruifeng]

I just test the modification in pyspark repl, and record the updated places in the attachment in the ticket

[SparkQA]

Test build #110039 has finished for PR 25662 at commit dc1e65e.

• This patch fails RAT tests.
• This patch merges cleanly.
• This patch adds no public classes.

[zhengruifeng]

This PR is too large, I need to divide it into small ones in some way.