add tril op and  triu op

* add unbind op
unbind(tensor, dim=0):
说明：移除指定维后，返回一组数组，包含了沿着指定维切片后的各个切片。

tensor(Tensor) -- 输入Tensor
dim(int) -- 删除的维度

示例：
 Input = [[1,2],
           [3,4],
           [5,6]]
  axis = 0
  Output[0] = [1,2]
  Output[1] = [3,4]
  Output[2] = [5,6]

* delete invalid check inferface Ref & VectorRef, test=develop

* fix vector ref delete error, test=develop

* try the new check inferface, test=develop

* change all related code with new check macro, test=develop

* remove static assert, test=develop

* polish detail, test=develop

* skip coverage problem, test=develop

* add new check macro, test=develop

* solve the conflict of ops with the same name. test=develop

* Add the argmax op to API 2.0, test=develop

* Fix the compiler problem in arg_max op, test=develop

* Fix the import meesage in common_ops_import, test=develop

* Fix the default dtype of arg_min_max, test=develop

Update the new api ops of creation ops to the api 2.0

* Add allclose Op, and its function is analogous to numpy.allclose. It returns True if two tensors are elementwise equal within a tolerance.

* sin sqrt tanh atan add out, test=develop

* optimize doc, test=develop

* add dygraph test, test=develop

* add tdm child op

* add flip op

* add functional conv

* add test and doc for function convs, test=develop

* update ConvTransposeOp's InferShape and error message, test=develop

* use global conv cache; test=develop

* use singleton cache; test=develop

* fix format error; test=develop

* add cudnn helper header; test=develop

* fix header error; test=develop

* fix mac unitest; test=develop

* fix mac unitest; test=develop

* fix file format; test=develop

* fix include file error, test=develop

* remove kernel_configs_ in class ExecutionContext and kernel_configs_map_ in class OperatorWithKernel, test=develop

* fix test_elementwise_mul_op_dim, test=develop

* fix compile error, test=develop
Co-authored-by: Nphlrain <phliuhongyu@126.com>

elementwise function used before definition then failed in cuda 8, move it ahead.

* update linspace, equal operators to API 2.0, test=develop

* equal support higher performance CUDA kernel, test=develop

* update comment of equal&linspace operator, test=develop

* update comment of equal&linspace operator, test=develop

* improve elementwise performance.

* Add contiguous check, test=develop

* add op inout check macro, test=develop

* fix enforce_test, test=develop

This reverts commit 48144e40.

* Add unittest for transformer prediction in dygraph_to_static.

* fix bug in fill_constant api. 

* Make transpose support size 0. test=develop

* use global conv cache; test=develop

* use singleton cache; test=develop

* fix format error; test=develop

* add cudnn helper header; test=develop

* fix header error; test=develop

* fix mac unitest; test=develop

* fix mac unitest; test=develop

* fix file format; test=develop

* fix include file error, test=develop

* remove kernel_configs_ in class ExecutionContext and kernel_configs_map_ in class OperatorWithKernel, test=develop

* fix test_elementwise_mul_op_dim, test=develop
Co-authored-by: Nphlrain <phliuhongyu@126.com>

* Delete is_test from activation operators
test=develop

* Revent unneeded changes
test=develop

* add inplace_abn_op. test=develop

* refine the error message of tensor_array_read_write Op

* add fleet pslib pull and push sparse op and push dense op
* test=develop

* Improve elementwise performance.

Elementwise performace is poor as walk into CommonGradBroadcastCUDA, add some new kernels for different data pattern.

* Add some cuda kernel to speedup common broadcast cases. test=develop

* Add more test cases and fix cuda kernel bug. test=develop

* Remove tests as cpu percision fails.test=develop

* Refine SplitDims, test=develop

* Change file mode, test=develop

* add dynamic plugin support.
test=develop

* change emb eltwise layernorm to math function
test=develop

* add emb eltwise layernorm
test=develop

* can run dynamic shape ernie
test=develop

* fix ci
test=develop

* add ut for trt ernie dynamic

test=develop

* refine dynamic shape c++ interface.
test=develop

* fix comments
test=develop

* fix comments
test=develop

* fix compile error in win gpu test=develop

* fix compile error in win gpu test=develop

* fix compile error in win gpu test=develop

* refine the transopose conv using v7 to choose algorithm

* add dequantize_log_op and make pyramid hash support int8 weight test=develop

* add unittest and update pyramid hash op test=develop

* remove paddle_enforce test=develop

* fix error message test=develop

* remove incorrent commit test=develop

* fix error message in log_dequantize test=develop

* change 2019 to 2020 test=develop

* remove useless check_grad test=develop