Cherry-pick bug fixes and samples from master to release/1.0

[codemzs]

Cherry pick for the below PRs from master branch, most of these PRs are related to samples and none related to API except for #3172 multi-column mapping API for normalizer estimator and #3291 that makes Prior trainer accept only Boolean type label column to make it consistent with every other binary trainer. I also did speak with @shauheen prior to cherry-picking and he agreed ideally these sample fixes should be in the release.

#3172
#3215
#3212
#3230
#3237
#3257
#3259
#3249
#3267
#3278
#3213
#3285
#3287
#3291
#3295

[shmoradims]

@codemzs please don't include code changes and documentation work in the same PR, if you want more people to be able to approve it. Now I can't approve this and you need to wait for Shauheen to approve.

[codemzs]

@shmoradims which commit are you referring to as documentation work here?

[shmoradims]

I'm approving all the changes related to API samples and xml documentation changes.

@shauheen please review changes in SimpleTrainer.cs as I didn't know if you want to take it or not.

[shmoradims]

[TomFinley]

So, let me point something out.

You have this (partial) commit history as part of this PR.

That first commit introduces an error that would make the project unable to build, until you fixed the error in the last commit. So if we accepted this as written, we'd have in our release/1.0 branch seven commits that did not build, thanks to this one cherry pick. Please rectify.

I am also unsure why a commit that describes itself as changing a namespace is introducing an entirely new sample. My best guess is that you made the commit for a sample that has not yet been cherry picked. That's potentially going to cause problems for whoever chooses cherry pick that sample later, dirties the commit history, etc.

Either way, it's something to be fixed.

Thanks!

[TomFinley]

Ah... here's another one I found. This cherry pick (commit id eb57d09) supposedly corresponds to #3287, but it includes content that was introduced originally in #3211, which should have been cherry picked separately. @shauheen for comment and direction on this.

Edit: I see there's already a cherry-pick PR for this in #3264 from @sfilipi. Happily that one has already been approved. Might be best for that to go in first.

[TomFinley]

I am also unsure why a commit that describes itself as changing a namespace is introducing an entirely new sample. My best guess is that you made the commit for a sample that has not yet been cherry picked. That's potentially going to cause problems for whoever chooses cherry pick that sample later, dirties the commit history, etc.

OK, so my guess was right. 😄 This sample, which you introduce as part of your commit 6593ed5 (supposedly corresponding to #3267), was actually introduced in #3099. Happily @zeahmed has a cherry pick PR #3240. If you want to structure things in this fashion, it might be best if you had that cherry pick go in before yours.

[shauheen]

All done as part of #3333 . Thanks @codemzs .