[Profiling tool]: Add the event to record the timeline. 

Add design documentation for profiling tool.

* Addign document for fluid split_lod_tensor

* Adding document for fluid merge_lod_tensor

Add doc for max_sequence_len.

del some DEPS in CMakeLists.txt  (merge Detection output op)

Add doc for lod_rank_table.

* "fix data transform"

* "data transformer"

* "add device pool"

* "add test"

* "fix ci"

* "fix datalayout implementation "

* "fix based on comment"

add Detection output op for SSD

Add backward doc and comments

Feature/rnn gradient check

Refine the activation type in the GRU operator related

Refine cos-sim-op

Adam selectedrows and scatter functors

update alexnet inference benchmark data on MKL

fix inference crash of alexnet benchmark