* Add a fake_quantize_op, which quantize an input tensor to a tensor with lower bits.

* Add region proposal network (RPN) target assign operator and Python API for Faster-RCNN.

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

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

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.

* move to platform

* "move init from framework to platform"

* "remove used init"

* "fix ci"

* "fix ci"

* "fix generic"

* "fix ci"

* "fix ci"

* "fix ci"

* "disable fragile test"

* Shrink reshape_op library size
* User can register a standard C++ functor as a op kernel

* Add anchor generator operator for Faster-RCNN.
* Add unittest testing.
* Add Python API.

This reverts commit bc28cf61.

* small AverageOptimizer enhance.

* clean

* clean

* Add fill_zeros_array op. This op is used for zero-filling an LoDTensorArray.

* merge fill_zeros_array_op with fill_zeros_like_op

* add unit_test for fill_zeros_like for array