add pull gpups sparse op

* updates the check_file_diff_approvals for allocation refactor, test=develop

* fix a bug, test=develop

* ipu add dockerfile

* resolve comments

* add test for conv_elementwise_add2_act_fuse_pass and conv_elementwise_add_act_fuse_pass

* Add conv_eltwiseadd_bn_fuse_pass test and fix test_conv_elementwise_addX_act_fuse_pass

* add tests for conv_act_mkldnn_fuse_pass

* add test for conv_bias_mkldnn_fuse_pass

* update code

* add conv_act_mkldnn_fuse_pass for relu, relu6, swish, leaky_relu

* update test

* update

* update bug

* update

* update pattern_detector

* fix test_conv_eltwiseadd_bn_fuse_pass

* add diff display notest;test=windows_ci_inference

* fix

* remove test_conv_act_mkldnn_fuse_pass.py

* ifix

* add layer_norm_fuse_pass test case

* restore cmakelist code

* Merge branch 'develop' into layer_norm_fuse_pass

* Merge branch 'develop' into layer_norm_fuse_pass

* add bad case test

* make some test higher paralle level

* modify slightly acoodring ci test:40100-40107

* modify slightly acoodring ci test:40124-40134

* modify slightly acoodring ci test:40138-40150

* modify slightly acoodring ci test:40153-40165

* modify slightly acoodring ci test:40153-40165

* add fc_elementwise_layernorm_fuse_pass

* fix name conflictn

* rebuild CI

* fix Ran Programs=0 bug

* make some non_parallel unittest parallel execute

* delete duplicate ut

* add test_unsqueeze2_eltwise_fuse_pass

* fix name conflictn

* rebuild CI

* test inference_api_test when run in windows-inference ci

* test if test failed the code run correctly

* put the failed test back

* refine structure for cuda and rocm

* update

* update

* update

* update

* update check_op_desc to support op_version and op_compat.

* modify for wincheck-inference case

* modify according to zhouwei's comment

* open with_trt and block failed unittests in windows

* test

Fix op-benchmark CI

* fix ut retry

* for pure fp16

* opt topk

* modify for wincheck-inference case

* modify according to zhouwei's comment

* modify according to zhouwei's comment 2nd time
Co-authored-by: Nzhangkaihuo <zhangkaihuo@baidu.com>

* Add transfer_layout/dtype op

* clean useless codes

* fix unused var

* add optest in white.txt

* split into data_transfer.cc

* fix cmake

* modify according reviewer comment

* replace cast_op with transfer_dtype_op

* modify unittest's parallel num

* modify unittest's parallel num

Update OP-benchamrk CI scripts 

* add benchmark