OneHotEncodingTransformer producing all zeros

[fealho]

• CTGAN version: 0.2.2.dev2

Currently, OneHotEncodingTransformer.transform may silently return vectors with all zeros, instead of raising an error. This happens whenever we try to transform data that we haven't seen before:

ohet = OneHotEncodingTransformer()
data = pd.Series(['a', 'b', 'c'])
ohet.fit(data)
ohet.transform(['d'])  #produces [0, 0, 0]           

I believe an error should be thrown is such situations, since it is not a one hot vector.