* update assign_pos

* update attention_lstm

* update barrier

* update batch_fc

* update beam_search

* update beam_search_decode

* update bilateral_slice

* fix bug

* Handle Structure kernel for InterpreterCore::RunOperator

* fix bug

* fix rocm compile

* fix rocm compile

* Revert "fix rocm compile"

* test

* revert test and update cmake

---------
Co-authored-by: Nchenruibiao <chenruibiao@baidu.com>

* add batch fc op, test=develop

* add batch_fc_op, test=develop

* fix untest, test=develop

* rm check_dygraph, test=develop

* fix comment, test=develop

* fix comment, test=develop

test=develop

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`

* Fix documentation

* Address review comments

* conv_shift_op: initial implementation using Eigen

Limitations:
- both gradient outputs must be specified and are always computed
- explicit for loops => could be optimized in various ways
  (e.g., different memory layout)

* conv shift - gradient fixes

fix case when not all output gradients desired

* conv shift: minor cleanup

* conv shift - more minor cleanup

* conv shift: clean up & initial GPU implementation

* fix rebase issue

* Make `PADDLE_ENFORCE_EQ` supports custom class, like DDim

* add NDEBUG switch to PADDLE_ENFORCE

* Free will be added soon