[SPARK-20862][MLLIB][PYTHON] Avoid passing float to ndarray.reshape in LogisticRegressionModel

## What changes were proposed in this pull request?

Fixed TypeError with python3 and numpy 1.12.1. Numpy's `reshape` no longer takes floats as arguments as of 1.12. Also, python3 uses float division for `/`, we should be using `//` to ensure that `_dataWithBiasSize` doesn't get set to a float.

## How was this patch tested?

Existing tests run using python3 and numpy 1.12.

Author: Bago Amirbekian <[email protected]>

Closes #18081 from MrBago/BF-py3floatbug.

(cherry picked from commit bc66a77)
Signed-off-by: Yanbo Liang <[email protected]>

Author: MrBago

python/pyspark/mllib/classification.py

@@ -171,7 +171,7 @@ def __init__(self, weights, intercept, numFeatures, numClasses):
             self._dataWithBiasSize = None
             self._weightsMatrix = None
         else:
-            self._dataWithBiasSize = self._coeff.size / (self._numClasses - 1)
+            self._dataWithBiasSize = self._coeff.size // (self._numClasses - 1)
             self._weightsMatrix = self._coeff.toArray().reshape(self._numClasses - 1,
                                                                 self._dataWithBiasSize)