* 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

* add lod_tensor util and modify pybind

* refind pybind LoDTensor API and modify LoDTensor and DataFeeder test

* fix test error

* fix detection map op test

* fix reorder_lod_tensor test

* fix seq_concat_op

* fix chunk evel op test

* fix target assign op

* fix warp ctc op

* address comments step 1: reverse reset_lod op

* step 2: modify op test

* add warning message

* remove has_valid_lod

* add back has_valid_lod

* address comments

* add exception catching trial

* 1. Create buddy allocator in each places before NcclBcast the variables
2. Check the memory usage of ALL gpus rather than the first one

* 1. Make NCCLGroupGuard guards only the ncclBcast part, which avoid ncclGroupEnd blocking the exception throwing
2. NOTE the usage of NCCLGroupGuard

* Remove the memory usage check of gpus

* Fix code style

* wip

* use executor.complete to end trainer

* fix build

* fix build with distribute off

* fix typo

* fix cmake typo

* fix build

* "add inplace attribute"

* "register inplace attribute"

* "change se-next model for memory-reuse"

* "fix typo"

* repick

* fix merge conflict

* "fix stupid error"

replace use_event with use_cuda, because use_event means the program running with CUDA, so use_cuda maybe more intuitive.

* fix dist train error

* update by comment

* "split into multiple .ccl"

* "refine file structure"

* "refine files"

* "remove the cmakelist"

* "fix typo"

* "fix typo"

* fix ci

* "init benchmark ops"

* "untrack outputs"

* "delete some usused code"

* "benchmark"

* "fix ci"

* "fix op test"

* "fix uint16 missing"

* "fix ci"

* "follow comments"

* "fix ci"

* "follow comments"

* "conficts. merge develop branch"

* repick

* "merge develop branch"

* Add MKLDNN layout support in Paddle

Add MKLDNN layout in Paddle so that MKLDNN friendly memory layout
can be used in MKLDNN enabled OP kernel. Before this commit, NCHW
is hardcode to be used in all MKLDNN op kernels. As a result,
non-optimized execution path is selected in MKLDNN primitive which
bring worse performance.
Besides framework change, three MKLDNN OP kernels were updated
for using new MKLDNN layout. They are conv/pool2d/batch_norm.
Other MKLDNN OP kernels need be also updated in similar way to
achieve best performance.

* Add MKLDNN layout support in activation OP

* Don't populate layout from input to output when kMKLDNN in

* Refine pool mkldnn op kernel

* MKLDNN layout

* Remove the inferitance from tensor file

* MKLDNN layout: refactoring

* Remove additional #define to register new operator

* Prepare mkldnn tests to work with layout