Remove Pearson correlation score based feature filtering for random effects

[ashelkovnykov]

No description provided.

[ashelkovnykov]

Updated for latest master.
@cmjiang @lguo Requesting review

Is Photon ML being published to an external artifact repository now? Which files need to be updated before a merge?

[cmjiang]

@ashelkovnykov The migration to external artifact repo is done. Please update release-bintray.gradle for new versions and CHANGELOG.md for new changes.

[ashelkovnykov]

@cmjiang
Does each change necessarily publish a new jar? What if we want to have multiple GitHub PRs condensed into one release, what should I do for that case? Asking because all five of my "ready to go" PRs should technically be a new major version, but I don't want to suddenly take us to version 25 of Photon ML. Thank you

[cmjiang]

@ashelkovnykov Good question. We don't need to push each change to a new jar. And we definitely don't want to suddenly bump 5 major version up. However the problem is that the version is explicitly pinned in the release-bintray.gradle. I can't think of a neat solution for now. @lguo @yunboouyang Any idea?

[cmjiang]

Since we don't have to push each change to a new jar, we can just keep the same version number for those PRs without new jars. The problem is that the same version number may have different versions of code.

[ashelkovnykov]

Forgot to remove the numFeaturesToSamplesRatioUpperBound from the data configuration