A bunch of doc issues

[ZhitingHu]

• TransformerDecoder.forward: where does self.training come from?
  

  texar-pytorch/texar/torch/modules/decoders/transformer_decoders.py

  Lines 448 to 449 in d17d502

  	infer_mode (optional): If not `None`, overrides mode given by
  	:attr:`self.training`.

• All arguments should say their types explicitly in the docstring. E.g., what is the type of infer_mode? The method signature is too dense to read for the type
  

  texar-pytorch/texar/torch/modules/decoders/transformer_decoders.py

  Lines 448 to 449 in d17d502

  	infer_mode (optional): If not `None`, overrides mode given by
  	:attr:`self.training`.

• TransformerDecoder.forward: start_tokens and end_token are mentioned but they are not explained in the argument docstring, and there is not explicit pointer that tells users where they can find the explanation.
  

  texar-pytorch/texar/torch/modules/decoders/transformer_decoders.py

  Line 353 in d17d502

  Arguments :attr:`(start_tokens, end_token)` are

• RNNDecoderBase.forward: An ordinary user (not developer) does not need to know dynamic_decode. It's confusing say Please refer to .. here. Say explicitly that it's needed only when one wants to over-write forward .. Also, we are using our own dynamic_decode -- should refer to here instead of tf.contrib.seq2seq.dynamic_decode
  

  texar-pytorch/texar/torch/modules/decoders/rnn_decoder_base.py

  Lines 104 to 107 in d17d502

  	Implementation calls :meth:`initialize` once and :meth:`step`
  	repeatedly on the decoder object. Please refer to
  	`tf.contrib.seq2seq.dynamic_decode
  	<https://www.tensorflow.org/api_docs/python/tf/contrib/seq2seq/dynamic_decode>`_.

• XLNetDecoder.forward: Is output just a list of sampled tokens or XLNetDecoderOutput?
  

  texar-pytorch/texar/torch/modules/decoders/xlnet_decoder.py

  Lines 318 to 320 in d17d502

  	:returns: A tuple of `(output, new_memory)`:
  	- **`output`**: The sampled tokens as a list of integers.
  	- **`new_memory`**: The memory of the sampled tokens.

• "helper" should be "helper_type"
  

  texar-pytorch/texar/torch/modules/decoders/xlnet_decoder.py

  Lines 313 to 314 in d17d502

  	helper: Type (or name of the type) of any sub-class of
  	:class:`~texar.torch.modules.Helper`.

[ZhitingHu]

• DecoderBase should be exposed in readthedocs, cuz a few important methods in it are used by users, like:
  

  texar-pytorch/texar/torch/modules/decoders/decoder_base.py

  Lines 144 to 150 in d17d502

  	def create_helper(self, *,
  	decoding_strategy: Optional[str] = None,
  	start_tokens: Optional[torch.LongTensor] = None,
  	end_token: Optional[int] = None,
  	softmax_temperature: Optional[float] = None,
  	infer_mode: Optional[bool] = None,
  	**kwargs) -> Helper: