test=develop

* fix error info for transpose sequence_conv_pool max_sequence_len sequence_erase, test=develop

* fix error info for transpose sequence_conv_pool max_sequence_len sequence_erase, test=develop

* update modify, test=develop

* update modify, test=develop

* fixed some modifications, test=develop

fix scatter and scatter_nd op error message

* update err info, test=develop

* update, test=develop

* polish some sl api error message, test=develop

* polish python input check of stride slice, test=develop

* fix unittest bugs, test=develop

test=develop

* refine Error message of modified_huber_loss, test=develop

* refine Error message style, test=develop

* enhance rank_loss error message, test=develop

* enhance similarity_focus error message, test=develop

* enhance squeeze error message, test=develop

* add scale_x_y for yolo_box, test=develop

* refine eps in iou_similarity, test=develop

* fix error log: resahpe, range, reverse.

* fix fsp_op error message,test=develop

* test=develop add linear interpolate operator

* Remove the error in the elementwise op, use the backup mode to calculate

test=develop

* add new macro BOOST_GET_SAFELY & unittests, test=develop

* add different macro type, test=develop

* fix get macro type in executor, test=develop

* four macro part change backup

* using one macro for all case, test=develop

* revert attribute change, test=develop

* change to three func to solve gcc4.8 bug, test=develop

* polish some details, test=develop

CreateStepScopes in recurrent op also clears scopes, which can cause segmentation fault un multi-threading. We add a lock in this PR but it may slow the computation process. We will fix in another way in next PR.

* support to train in static

* support to independent decorator

* remove in_dygraph_mode condition in ProgramTranslator

* fix import param_guard and add train/eval test=develop

* Modify into ShareVarsFromScope and rm __all__ in partial_program test=develop

1. To make ProgramTranslator to support `assert` grammar, this PR adds `assert` python API and C++ code. 

2. Fix a bug: graph_pattern_detector.h #include <gtest/gtest_prod.h> but didn't declared dependency at CMakeLists, which can cause single build failure.

3. Refactoring `Formatter` in print_op to make it reusable and reuse the formatter to print in assert op.

test=develop
test=win_gpu

* API/OP (margin_rank_loss, nce, row_conv, glu, positive_negative_pair) error message enhancement, test=develop

* add unittest for glu, test=develop

* Replace dygraph_to_static_func with @declarative or program_translator.get_func in test_list.py

* Add comments in ConditionalBlock.

* Support list pop last item. 

* Support pop the i-th item. 

* Support an empty tensor array as Input in assign op and set the kernel type is float.

* improve activation ops performance, test=develop

* use 32bit only GPU computation, test=develop

* Support LoDTensorArray for slice op.
* Support read elements of list in dygraph_to_static
* Fix infershape add test for infershape.
* Support Tensor for Attr(starts) and Attr(ends). 
* Use new interfaces in VarTypeInference. 

fix kron_op: when only one input needs gradient

* enhance add_position_encoding error message, test=develop

* enhance warpctc & scaled_dot_product_attention error message, test=develop

* modified error message and ctest of scaled_dot_product_attention, test=develop

* remove double registery for pyramid_hash op

* add noavx_axpy and noavx_axpy_noadd

test=develop

* add print transformer & unify print format, test=develop

* remove using of dygraph_to_static_func, test=develop

* remove python stdout capture, test=develop

* fix compatibility problems for PY2, test=develop

* fix detail error, test=develop

* fix type analysis bug, test=develop

* fix print tuple compatible error in PY2, test=develop

* replace get_func to declarative, test=develop

* fix detail bug, test=develop

* fix some detail problems, test=develop

* change visit_call in print transformer, test=develop

* add trace op, test=develop

* Optimized the implementation of trace op, test=develop

* fix a bug of include in trace_op.h, test=develop

* move trace API from creation to math, test=develop

* modified en doc. test=develop

* add complex trace api

* add complex sum api, test=develop

* modified en doc of complex sum and trace, test=develop

* modified doc and trace API, test=develop

* modified en doc of trace and sum, test=develop

* modified comment in complex kron API, test=develop

* OP Should Not Have Unused Input, test=develop

* add GetExpectedKernelType, test=develop