* Add trainable_statistics in attr for batch_norm
* Unifying behavior of dynamic graph and static graph

test=develop

* change int to floor, test=develop

* fix unittest, test=develop

* test=develop

* support the head number == 1
test=develop

* fix slice op error.
test=develop

* Make batch_float_feasigns & batch_uint64_feasigns as member variable

* update name generator, test=develop

* use c++ unique name generator, test=develop

Fix the dygraph mode in matmul, add the support in Linear Op

* add the thread_local_allocator, test=develop

* refactor the thread_local_allocator, test=develop

* provides option setting strategy, test=develop

* add DataLoader, Dataset, BatchSampler

* add clamp api, test=develop

* OP(fusion_gru) error message enhancement. test=develop

* refine code, test=develop

* Refine inout log, test=develop

* Refine description, test=develop

* Optimize the error messages of paddle CUDA API, test=develop

* fix the error messages of paddle CUDA API, test=develop

* Refactoring PADDLE_ENFORCE_CUDA_SUCCESS, and apply to curand/cudnn/cublas/NCCL,test=develop

* remove build_ex_string,test=develop

* merge conflict,test=develop

* test=develop, add seed arguemnt of randint API

* add diag_embed op, test=develop

* add TestCase of diag_embed API

* Modified diag embed python API teastcase from dygraph to static graph, test=develop

* delete useless log and trigger ci, test=develop

* modified float16 of diag_embed, test=develop

* modified en doc of diag_embed

* trigger ci, test=develop

* add fp16 in dtype check of python API, test=develop

* modified __init__ and fix a big, test=develop

* modified a test bug of test_bicubic_interp_op and test_trilinear_interp_op, test=develop

* modified to use one kernel on cpu and cuda, test=develop

* enhance pad.* error message, test=develop

* optimize rank_attention, test=develop

* use the paddle memory pool, test=develop

* set max size, test=develop

* limit the max size, test=develop

* fix the head of cu, test=develop

* add AsDispensable, test=develop

restrict block num of layer_norm_grad cuda kernel to 128, test=develop

* refine ernie trt dynamic shape support
1. add slice op converter
2. add emb eltwise layernorm fp16 support
test=develop

* fix dynamic shape test ut
test=develop

* fix comments.
test=develop

* fix comments
test=develop