Add key_padding_mask and attn_mask in sparse_attention Api

1.Key padding mask is a tensor with dimensions [batch_size, seq_len], and attention mask is a tensor with dimensions [seq_len, seq_len]. The data types of the two masks are consistent with Q, K, and V, which are float32 or float64. If the value in Mask is 0, it means that the position needs to be masked.

2.The changed files are mainly paddle/fluid/operators/sparse_attention_op.cu and python/paddle/fluid/tests/unittests/test_sparse_attention_op.py. sparse_attention has three parts: sddmm, softmax, and dsd. Adding the mask operation only needs to modify the softmax. It has no effect on the other two parts. In addition, in order to test the mask function, related tests has been added.

* Add the transformop parameter in TensorReduceFunctorImpl

* Reduce reshape kernel functions in pten

* delete notes

* fix bugs when compile

* modify register name

* fix compile bugs

* unify device context entrance

* move all_context include to header

* polish cmake relay for device_context

* fix npu compile failed

* fix npu compile failed

* revert part of change

* Revert "Revert "pylayer support tuple/list type args (#37727)" (#37956)"

This reverts commit d848ff04.

* move check args,kwargs before forward execute

* Add float16 type for scatter op.

* Add fp16 test for scatter op.

* Add int and int64 support for scatter_grad on gpu.

* Add int and int64 for check_variable_and_dtype routine.

* Minors.

* Code format.

* Rearranged Eager AutoCodeGen directory structure

* Removed USE_OP in Eager AutoCodeGen

* Enabled generation for Operators without Grad/Inputs/Outputs

* Resolved operators without input

* Fixed merge conflicts

* Enabled Eager AutoCodeGen for 10+ more operators

* Refactored Eager AutoCodeGen with more organized helper objects

* Enabled Eager AutoCodeGen for operators with multiple OpBases

* Adjusted Eager AutoCodeGen to Enable Passing Output Tensor as Input Argument

* Handled Dispensable Inputs/Outputs in Eager AutoCodeGen

* Enabled Eager AutoCodeGen for All Existing Operators & Possible Future Operators

* Fixed CI issues

* update mkldnn scale_matmul fuse pass ut

* update mkldnn scale_matmul_fuse_pass ut

* add mkldnn conv3d_bias_mkldnn_fuse_pass ut

* update conv3d_bias_mkldnn_fuse_pass ut

* disable conv3d_bias_mkldnn_fuse_pass

test=document_fix

* fix bugs in Translated layer when change train/eval

* fix python converage

* fix error when tensor_shape_transformer. Before in stmt like `if len(paddle.shape(x)[0]) > 0`, `paddle` will be used as a variable

* handle other call like `fluid.layers.mean` and `fluid.layers.shape`

* add unit test

* add ipu_inference

* resovle commments

* resolve comments

* add EnableIpu introduction

* rm line

* restore npu update

* add ernie and resnet50 test

* fix copyright time
Co-authored-by: Nyaozhixin <522190855@qq.com>

* refine test

* add download_program target

* update ut code

* refine code

* disable profiler

* add comments

* refine cmake

* skip coverage ci

* add hinge_embedding_loss

* fix test_API

* test_API succeed

* add English doc

* fixed using of expired fluid api

* fix doc

* fix doc and rm python/paddle/fluid/layers/loss.py

* get raw python/paddle/fluid/layers/loss.py back

* fix Examples bug in English doc

* unique -> flatten

* fix api code

* fix English doc

* fix functional loss English doc

* fix Example doc

* .numpy() -> paddle.unique()

* fix unique

* fix label_item_set

* modified judgment equation

* Got a beautiful loss equation

* use paddle.to_tensor

* fix loss and add static check

* fix loss and add static check

* delta -> margin

* update mkldnn conv_concat_relu_mkldnn_fuse_pass ut

* update conv_concat_relu_mkldnn_fuse_pass ut

* restrict conv2d data_format in conv_concat_relu_mkldnn_fuse_pass

* remove bf16

* remove comments

* remove wrong return

* fix UT

test=document_fix

* add new API:paddle.movedim/moveaxis

* add new API:paddle.movedim/moveaxis

* add new API:add new API:paddle.movedim/moveaxis

* fix comment

* fix comment

As the title.

* replace with pten kernel in cast cuda compute and remove unused codes

* rm unused header file

* replace CastCUDAOpKernel with CastOpKernel

* Rearranged Eager AutoCodeGen directory structure

* Removed USE_OP in Eager AutoCodeGen

* Enabled generation for Operators without Grad/Inputs/Outputs

* Resolved operators without input

* Fixed merge conflicts

* Enabled Eager AutoCodeGen for 10+ more operators

* Refactored Eager AutoCodeGen with more organized helper objects

* Enabled Eager AutoCodeGen for operators with multiple OpBases

* Adjusted Eager AutoCodeGen to Enable Passing Output Tensor as Input Argument

* Handled Dispensable Inputs/Outputs in Eager AutoCodeGen

* Adjusted function generation/call between Python-C API & Dygraph API

* Synchronized auto-generated Python-C API with Dygraph Forward Functions

* Added safe_initialized interface to EagerTensor for use in processing dispensable inputs

* Enhance error msg in paddle.assign in static mode

* fix unittest

* add map_matmul passes to quant2_int8_mkldnn_pass

* fix fc+act fuse (activation scale)

* ci fix, c++17 structured bindings not available

* fix ci static check

* fix bug of set_value op

* fix BumpInplaceVersion

* polish some comments

* revert change of full_like

* update Planner

* update unitest

* update PlanSpace

* update PlanSpace

* modify set_grad_var_shape

* update code style

* polish register marco

* resolve compile failed

* revert needless change

* revert eager related change

* revert eager related change

* change register marco name

* polish deetails

* add conv_gelu_mkldnn_fuse_pass

* add post ops