Add loadable attribute for LinearMulticlassModelParameters

[Ivanidzo4ka]

fixes #3209.

[codecov]

Codecov Report

Merging #3217 into master will increase coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #3217      +/-   ##
==========================================
+ Coverage    72.6%   72.62%   +0.01%     
==========================================
  Files         807      807              
  Lines      145077   145080       +3     
  Branches    16213    16213              
==========================================
+ Hits       105337   105366      +29     
+ Misses      35322    35297      -25     
+ Partials     4418     4417       -1

Flag	Coverage Δ
#Debug	72.62% <100%> (+0.01%)	⬆️
#production	68.17% <ø> (+0.02%)	⬆️
#test	88.92% <100%> (ø)	⬆️

Impacted Files	Coverage Δ
...LogisticRegression/MulticlassLogisticRegression.cs	67.61% <ø> (+1.74%)	⬆️
.../Microsoft.ML.Tests/TrainerEstimators/SdcaTests.cs	97.31% <100%> (+0.05%)	⬆️
src/Microsoft.ML.Maml/MAML.cs	24.75% <0%> (-1.46%)	⬇️
...ML.Transforms/Text/StopWordsRemovingTransformer.cs	86.26% <0%> (+0.15%)	⬆️
...oft.ML.StandardTrainers/Standard/SdcaMulticlass.cs	91.12% <0%> (+1.02%)	⬆️
...oft.ML.StandardTrainers/StandardTrainersCatalog.cs	92.34% <0%> (+3.27%)	⬆️
...c/Microsoft.ML.FastTree/Utils/ThreadTaskManager.cs	100% <0%> (+20.51%)	⬆️

[abgoswam]

No Prob [](start = 19, length = 7)

should we say "no calibration" or "raw scores" or "logits" ?

right now it could be interpreted as "no problem" :)

[Ivanidzo4ka]

I'm afraid we have limit in 26 chars for this string, and no calibration doesn't fit.
I'm fine with Raw score, but I would prefer to @wschin comment on this one, since he was one who split them apart

[wschin]

nit: Multiclass LR No Cali",

[abgoswam]