[Paddle-Inference] support preln-ernie: add preln_emb_eltwise_layernorm_op, preln_skip_layernorm_op (#39570)

* support preln_ernie: add preln_emb_eltwise_layernorm_op, preln_skip_layernorm_op

* support preln_ernie: add preln_emb_eltwise_layernorm_op, preln_skip_layernorm_op

* remove reshape and elementwise_add registry

* delete code

* fix bugs when run ci ut

* remove log

* fix bugs when run unit test

* fix bugs when run unit test

* fix bugs when run cinn

* fix bugs when run ci-mac-python3

* fix compile bugs

* fix compile bugs

* fix compile bugs

* fix bugs when run kunlun

* fix bugs when compile

* update code according comment

* Replace GeLU plugin with TRT built-in layers for approximate GeLU

* Add TensorRT built-in layer for nonapproximate GeLU

* support ernie quant model with interleaved

* support ernie quant model with interleaved

* support ernie quant model with interleaved

* support ernie quant model with interleaved

* support ernie quant model with interleaved

* support ernie quant model with interleaved

* support ernie quant model with interleaved

* rebuild matmul pass: trt and gpu_cpu

* rebuild matmul pass: trt and gpu_cpu

* rebuild matmul pass: trt and gpu_cpu

* rebuild matmul pass: trt and gpu_cpu

* slice

* shuffle pass enhancement

* shuffle channel pass

* add ut

* timeout fix

* makefile fix

* Paddle-Inference:fix_concat_slice

* Paddle-Inference:fix_concat_slice

* Paddle-Inference:fix_concat_slice

* Paddle-Inference:fix_concat_slice

* [Paddle-Inference]: fix concat slice

* [Paddle-Inference]: fix concat slice

* [Paddle-Inference]: fix concat slice

* [inference] update convert reduce op&ut,test=develop

* update

* update

* update

* add int32 support

* add int32 support

* add comments

* trt < 7.0 do not support int32

* test=develop

* update

* test=develop

* Merged LoDTensor with Tensor,test=allcases

* Patched python level LoDTensor

* Patched python level LoDTensor

* Merge Tensor into DenseTensor

* Fixed namespace issues,test=allcases

* Fixed merge issues

* Fixed inference issues

* Fixed NPU test issues

* Fixed merge issues

* fix trt convert conv2d skip

* fix trt convert conv2d skip

* modify params check

* correct compile

* develop test

* throw

* ne

* wrong cnt

* add trt_convert_flatten_contiguous_rang op

* trt version >7,support trt_convert_flatten_contiguous_rang

* trt version >7,support trt_convert_flatten_contiguous_rang

* trt version >7,support trt_convert_flatten_contiguous_rang

* test cast add trt version >=7 skip

* add Paddle Trt config: with_interleaved

roi_align aligned supported

* bug fix

* remove blank

* dynamic shape clone supported

* Inspect the information inside a TRT engine.

* Follow up the google code style.

* Fix code error.

* add restriction on plugin supportsFormat to eliminate errors from TensorRT8

* ernie-varlen is only supported on architecture >= sm75

* fix_ele_convert: IElementWiseLayer can broadcast

* fix_ele_convert

fix: when ceil_model==true && Padding_algo!=SAME, (x-size)/stride != int, this convert is wrong (#37929)

* fix

* Update stack_op_plugin.cu

comments

* refine structure for cuda and rocm

* update

* update

* update

* update

* support data_format='NHWC' for prelu channel mode

* matmul_convert_int8

* matmul_convert_int8

* matmulconvert_int8

* Matmul_int8_convert: tensor*tensor

* Matmul_int8_convert: tensor*tensor

* Matmul_int8_convert: tensor*tensor

* modify code about fp16 of dcnv2 trt

* fix_nearest

* fix_nearest

* fix_nearest

* fix_nearest

* remove input dim check of activation in op_teller

* remove input dim check of concat in op_teller

* remove input dim check of clip in op_teller

* remove input dim check of scale in op_teller

* remove input dim check in op_teller

* update attr check of slice in op_teller

* fix_qkv_plugin: half_scale

* [Paddle-Inference] fix_qkv_plugin: fix half scale