Add none layers api doc

* Make the CUDA kernel of concat correct and fix unit tests.

Polish profiler Python API.

fix mkldnn compile issue

Refine detection_map doc.

add ftrl optimizer

* Add sub_blocks of lr_decay_op to pserver_prog after distribute_transpiler

* Remove unused logs and logics

* 1. Add ops to new block (considering the nested block condition)
2. Follow the original hierarchy of blocks
3. Change the function's name and remove debug lines

refine the initial cpu memory flag for mkldnn

concat support data as input

add doc for inference_transpiler

MKLDNN layout: Support for activation operator

* Remove tape

* remove tape in cmake

* fix CI