* 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.

Change patch version to str

Replace {} to + in cmake regex match

Change the PADDLE_VERSION in develop branch to latest

* use -rpath to fix libmklml_intel.so not found

* only remain $ORIGIN/../libs in rpath of core.so

* test

* test

* add rpath of libmkldnn.so.0

* check return value of os.system

* remove check return value of patchelf

This reverts commit 36d9e4be.

* add note when import core.so error

* update

* make transpiler test reliable

* add more

* follow comments

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

This is part of big move of v2 codes to legacy sub dir.