[SPARK-45757][ML] Avoid re-computation of NNZ in Binarizer

[zhengruifeng]

What changes were proposed in this pull request?

1, compress vectors with given nnz in Binarizer;

2, rename internal function def compressed(nnz: Int): Vector to avoid ambiguous reference issue (vec.compressed.apply(nnz)) when there is no type hint

[error] /Users/ruifeng.zheng/Dev/spark/mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala:132:61: ambiguous reference to overloaded definition,
[error] both method compressed in trait Vector of type (nnz: Int): org.apache.spark.ml.linalg.Vector
[error] and  method compressed in trait Vector of type org.apache.spark.ml.linalg.Vector

Why are the changes needed?

nnz is known before compression

Does this PR introduce any user-facing change?

no

How was this patch tested?

ci

Was this patch authored or co-authored using generative AI tooling?

no

[dongjoon-hyun]

+1, LGTM.

[dongjoon-hyun]

Merged to master for Apache Spark 4.0.0.