Refactor whole data preprocessor for DS2 (re-design classes, re-organize dir, add augmentaion interfaces etc.).

1. Refactor data preprocessor with new added class AudioSegment, SpeechSegment, TextFeaturizer, AudioFeaturizer, SpeechFeaturizer.
2. Add data augmentation interfaces and class AugmentorBase, AugmentationPipeline, VolumnPerturbAugmentor etc..
3. Seperate normalizer's mean and std computing from training, by adding FeatureNormalizer and a seperate tool compute_mean_std.py.
4. Re-organize directory.

Support variable input batch and SortaGrad.

add tools for model conversion from Caffe to PaddlePaddle

update the text generation demo.

Update librispeech.py for DeepSpeech2.

Add the NCE example.

Add model resuming for DeepSpeech2 trainer.

Summary:
1. Add manifest line check.
2. Avoid re-unpacking if unpacked data already exists.
3. Add full_download (download all 7 sub-datasets of LibriSpeech).

fix ctr html error that makes CI fail.

convert markdown into html.

Refactor decoder interfaces and add ./data directory.

Add audio data provider and a simplified DeepSpeech2 model configuration.

2. Fix incorrect batch-norm usage in RNN.
3. Fix overlapping train/dev/test manfests.
4. Update README.md and requirements.txt.
5. Expose more arguments to users in argparser.
6. Update all other details.

update readme.

update README.

add the CTR demo.