1. Make while_op accept GPU conditional data
2. Add more complex test cases for while_loop API

test=develop

* Remove self-set accuracy parameters of op tests: max_relative_error
test=develop

* fix errors
test=develop

*use dtype to determine whether check_grade is needed, and delete useless class

The fixed bugs:

1. The condition sub-graph is not pruned
2. When backward graph is extremely simple, the whole backward ops are pruned.

test=develop

adds unary operator __neg__ for VarBase in dygraph mode, and for Variable in static graph mode.

* Remove self-set accuracy parameters of op tests: max_relative_error
test=develop

* Remove self-set accuracy parameters of op tests: max_relative_error
test=develop

* Remove self-set parameters of op tests: max_relative_error
test=develop

* Remove self-set accuracy parameters of op tests: max_relative_error
test=develop

* Remove self-set accuracy parameters of op tests
test=develop

* keep smaller parameters
test=develop

* increase example code of py_func, fix some wrong description of English API doc

test=develop

* Remove self-set accuracy parameters of op tests
test=develop

* add cast
test=develop

* Remove self-set accuracy parameters of op tests:atol
test=develop

* lrn_ngraph_op
test=develop

* Keep smaller parameters
test=develop

* remove build_once & name_scope (Conv2D)
test=develop

* fix unittest
test=develop

* Conv2DTranspose

* Conv3D & Conv3DTranspose
test=develop

* Pool2D & BatchNorm

* Embedding

* LayerNorm

* GRUUnit & NCE

* PRelu

* BilinearTensorProduct

* GroupNorm & SpectralNorm

* TreeConv
test=develop

* fix LayerNorm in transformer unnittest
test=develop

* disable LayerNorm or BatchNorm in multicard
test=develop

* refine Layer.create_parameter api
test=develop

* refine LayerNorm, remove begin_norm_axis param, add normed shape check
test=develop

* LayerNorm bug fix
test=develop

* Fix softmax cuda bug

* Refine multihead log and softmax logic

* yolo_box OP add Attr(clip_bbox). test=develop

- ngarph: elementwise_sub, elementwise_mul
- mkldnn: transpose, sum
- others: scatter_nd

test=develop

test=develop

* fix elementwise_pow bug on integer, test=develop

* use llrint to support elementwise_pow_grad, test=develop

* add some tests, test=develop

* revert grad functor, test=develop

* patch math method for varbase using auto-generated op functions, test=develop

* clean code that handles batch_size, test=develop

* follow comments, test=develop

* follow comments, test=develop

* code clean, test=develop

* add backward unittest test=develop

* add sub-branch in test_backward.py test=develop

* refine code, add comment test=develop

* reconstruct TestBackward Class test=develop

* fix typo of comment test=develop

* update op test framework

affine_grid, label_smooth, spectral_norm, warpctc,
nearest_interp, data_norm, match_matrix_tensor,
var_conv_2d, fused_embedding_seq_pool

test=develop

sequence_expand_as, squared_l2_distance, gather_nd,
center_loss, rank_loss, conv_shift, spp,
modified_huber_loss, smooth_l1_loss, multiplex,
sequence_softmax, nce, huber_loss, group_norm,
kldiv_loss, hinge_loss, expand_as

test=develop

* Add reshape int8 op

test=develop

* Change test to CPUPlace

test=develop

* Correct tests

test=develop