Enable more loss types for OpLinearRegression

[TuanNguyen27]

Since Spark 2.3.0, Huber loss is supported for SparkML's linear regression, which will be more robust against outliers.

[codecov]

Codecov Report

Merging #421 into master will increase coverage by 4.87%.
The diff coverage is 50%.

@@            Coverage Diff             @@
##           master     #421      +/-   ##
==========================================
+ Coverage   82.09%   86.96%   +4.87%     
==========================================
  Files         337      337              
  Lines       11078    11082       +4     
  Branches      597      588       -9     
==========================================
+ Hits         9094     9637     +543     
+ Misses       1984     1445     -539

Impacted Files	Coverage Δ
...op/stages/impl/regression/OpLinearRegression.scala	76.92% <50%> (-4.9%)	⬇️
...es/src/main/scala/com/salesforce/op/OpParams.scala	85.71% <0%> (-4.09%)	⬇️
...sforce/op/stages/impl/feature/Transmogrifier.scala	98.02% <0%> (+1.12%)	⬆️
...com/salesforce/op/utils/stages/FitStagesUtil.scala	94.73% <0%> (+1.31%)	⬆️
...ala/com/salesforce/op/features/types/package.scala	57.93% <0%> (+1.37%)	⬆️
.../scala/com/salesforce/op/dsl/RichTextFeature.scala	83.09% <0%> (+1.4%)	⬆️
.../main/scala/com/salesforce/op/OpWorkflowCore.scala	95.45% <0%> (+1.51%)	⬆️
...n/scala/com/salesforce/op/readers/DataReader.scala	95.23% <0%> (+1.58%)	⬆️
...orce/op/stages/impl/tuning/OpCrossValidation.scala	97.95% <0%> (+2.04%)	⬆️
...a/com/salesforce/op/filters/RawFeatureFilter.scala	92.97% <0%> (+2.16%)	⬆️
... and 32 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b67626f...72960e8. Read the comment docs.