* API/OP (Some SL API) error message enhancement  (#24441)

* polish some sl api error message, test=develop

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

* fix unittest bugs, test=develop

* fix error info for transpose sequence_conv_pool max_sequence_len sequ… (#24437)

* 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
Co-authored-by: NXing Wu <wuxing03@baidu.com>

* add lod check for sequence op, test=develop

* delete unnecessary check in expend op, test=develop

* Add the check of lod in sequence_softmax kernel.
test=develop

* Refine the comments.
test=develop

* wip simplify operator framework

* wip

* wip

* done test=develop

* clean test=develop

* fix test=develop

* fix deps test=develop

* fix cpu build test=develop

* fix tensorrt build test=develop

* fix tests test=develop

* fix test=develop

* fix cpu build test=develop

* refine seq_softmax_op

* fix seq_softmax

* use cub in seq_softmax

There are mainly following fixes:

- take `DeviceContext` as the template parameter of math functors and OpKernel instead of `Place`
- remove `eigen_device` interface in base class  `DeviceContext`
- remove `GetEigenDevice` interface in `ExecutionContext` and base class `DeviceContext`
- remove unused `platform::EigenDeviceConverter`
- rename `REGISTER_OP_GPU_KERNEL` to `REGISTER_OP_CUDA_KERNEL`
- rename `USE_GPU_ONLY_OP` to `USE_CUDA_ONLY_OP`

* Remove template parameter for Tensor methods

* Also check the type is correct when data()
* Simplize holder_

* Fix accuracy_op

* Register Code

* No POD instance is forbidden in global scope.

See
https://google.github.io/styleguide/cppguide.html#Static_and_Global_Variables

Make implement an operator less noisy.

* add inputs

* add ut for multiple inputs

* fix AddToLayer

* op_desc -> op_proto

* CreateArgumentOffsetMap -> CreateInOutOffsetMap

* move CreateInOutOffsetMap from OperatorBase to op registry

* arg_idxs_ -> in_out_idxs_

* Convert `op` --> `operators`
* Remove AddType in OpProtoMaker, because type is part of registry.
* Rename CPU_OR_GPU --> DEVICE_TYPE in registry macro.