Bugfix: pass through vocab/encoding params (#60)

This appears to be a typo. I think the correct behavior is that GoodTuring should take into account a vocab and encoding if given.

Author: Eric Ihli

numpy_ml/ngram/ngram.py

@@ -506,7 +506,7 @@ def train(self, corpus_fp, vocab=None, encoding=None):
             Specifies the text encoding for corpus. Common entries are 'utf-8',
             'utf-8-sig', 'utf-16'. Default is None.
         """
-        self._train(corpus_fp, vocab=None, encoding=None)
+        self._train(corpus_fp, vocab=vocab, encoding=encoding)
         self._calc_smoothed_counts()
 
     def log_prob(self, words, N):