[SPARK-18408][ML] API Improvements for LSH

## What changes were proposed in this pull request?

(1) Change output schema to `Array of Vector` instead of `Vectors`
(2) Use `numHashTables` as the dimension of Array
(3) Rename `RandomProjection` to `BucketedRandomProjectionLSH`, `MinHash` to `MinHashLSH`
(4) Make `randUnitVectors/randCoefficients` private
(5) Make Multi-Probe NN Search and `hashDistance` private for future discussion

Saved for future PRs:
(1) AND-amplification and `numHashFunctions` as the dimension of Vector are saved for a future PR.
(2) `hashDistance` and MultiProbe NN Search needs more discussion. The current implementation is just a backward compatible one.

## How was this patch tested?
Related unit tests are modified to make sure the performance of LSH are ensured, and the outputs of the APIs meets expectation.

Author: Yun Ni <yunn@uber.com>
Author: Yunni <Euler57721@gmail.com>

Closes #15874 from Yunni/SPARK-18408-yunn-api-improvements.