`LinearModel` error when detecting target column types for early stopping

[oleksiyskononenko]

The perfectly valid LinearModel fit call

from datatable import dt
from datatable.models import LinearModel

DT_X = dt.Frame([1, 2, 3])
DT_y = dt.Frame([True, False, True])

lm = LinearModel()
lm.fit(DT_X, DT_y, DT_X, DT_y)

errors with

TypeError: Training and validation target columns must have the same ltype, got: int and bool

The issue here is that the model didn't properly read the training target column type.