[SPARK-11763] [ML] Add save,load to LogisticRegression Estimator

[jkbradley]

Add save/load to LogisticRegression Estimator, and refactor tests a little to make it easier to add similar support to other Estimator, Model pairs.

Moved LogisticRegressionReader/Writer to within LogisticRegressionModel

CC: @mengxr

[jkbradley]

This is awkward, but type issues prevented me from using apply instead of get

[jkbradley]

Oops, I just realized I forgot to save the parent field in both this and Pipeline. Shall we just document that it does not get saved, and handle it for 1.7?

[SparkQA]

Test build #46032 has finished for PR 9749 at commit 528b5b1.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):\n * class BitSet(numBits: Int) extends Serializable\n

[mengxr]

We won't need this if we implement testDefaultReadWrite properly.

[mengxr]

LGTM except one minor issue commented inline. It should be okay to postpone saving parent to 1.7.

[jkbradley]

Thanks, updated!

[SparkQA]

Test build #46102 has finished for PR 9749 at commit 54a4885.

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

[mengxr]

We don't need pull in Path from Hadoop. This should work:

val subdir = new File(tempDir, subdirName)
val path = new File(subdir, uid).getPath

[jkbradley]

Rebased to pull PipelineSuite from master.
I removed the failing part of the PipelineSuite. It was not really needed since other tests already check that save/load works and that stages are saved to appropriate directories.

[SparkQA]

Test build #46110 has finished for PR 9749 at commit 1f61d86.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[mengxr]

Merged into master and branch-1.6. Thanks!