binning API now can return a factor column or an integer one

[nhanitvn]

Description and related tickets, documents

• That is done by adding toLabels params to allow the output column to be an integer one which contains zero-based indices instead of a factor column of range strings
• Refactor the code, move the implementation to ABinningHandler to avoid duplicated code in spark/jdbc BinningHandler
• Add convenient methods for binning to DDF while keeping the old one for backward compatible
• Add precision param for number of decimal digits to display in range strings (default is 2)
• Automatically search for appropriate precision value (20 tries from the specified precision value, default is 2) for nice labels
• Add validation for duplicated/non-monotonically-increasing breaks
• Binning by equal intervals and by equal frequency always use includeLowest = True despite any supplied value
• In case of binning by equal intervals, if min = max, the lower and upper bounds are extended by .1% as in Pandas and R to ensure that we can still map values to a range/value.
• Change EQUAlFREQ (lower-cased L) to EQUALFREQ (upper-cased L)

Reviewers: @hai-adatao @phvu @Huandao0812

Breaking changes & backward compatible issues

How to test

Describe how this PR is tested. In case manual testing is required, describe how to do so.

PR Progress

Make sure all checkboxes below are checked before merged

• [x] Branch is in format prefix/description (see this)
• [x] Merge check has no conflicts. PR checks passed.
• [ ] Code review is done.

[hai-adatao]

@nhanitvn can you help put all these tests in ddf-test/BinningHandlerBaseSuite, and then implement spark test suite instead? Check out JoinHandlerBaseSuite and JoinHandlerSuite for an example. Also, would appreciate if you take a look at the tests in ddf-test/BinningHandlerBaseSuite, they're all ignored, if they're not valid anymore please help remove it, or refactor it otherwise.

[hai-adatao]

Let's try to get this in 1.4.17