Merge branch 'complete_py_reader_python' of https://github.com/sneaxiy/Paddle into complete_py_reader_python

Merge branch 'complete_py_reader_python' of https://github.com/sneaxiy/Paddle into complete_py_reader_python

Merge branch 'complete_py_reader_python' of https://github.com/sneaxiy/Paddle into complete_py_reader_python

[WIP] Dev reader reset all

* Fix batch norm when only 1 elements in normzalize dimension during training.

Update hyperlinks in workflow_of_capi_cn.md

fix compile warning in inference related codes

pull from base

1. Fix bug when only true or false branch works.
2. Fix bug in unit testing.

Update version via git tag

Merge branch 'dev_reader_shutdown_start' of https://github.com/JiayiFeng/Paddle into dev_reader_shutdown_start

version it is staying on

workaround for no MKLDNN kernel

Update dist lookup table design doc

bugfix/libpaddle_inference_api.so double free

1. Shrink DoubleBufferReader's buffer size to 3.
2. Add BatchReader an option to discard leftover instances.
3. Fix a MultiPassReader bug on pass count.

1. Make PyReader and RandomDataGenerator inherited from FileReader.
2. Remove the memeber variable 'dims_' and realated checks in FileReader.

We can get endpoints of a reader chain.

release branch in the futrue

Merge branch 'complete_py_reader_python' of https://github.com/sneaxiy/Paddle into complete_py_reader_python

Adding corresponding hl_fini() for each hl_init() used