* heter bug

* format

* format

* add softmax_switch for softmax_with_cross_entropy_op, test=develop

* delete using EigenMatrix in softmax_with_cross_entropy_op.h, test=develop

* add REGISTER_OP_VERSION for softmax_switch attr of softmax_with_cross_entropy_op, test=develop

* remove the send/recv of tensor size, but users have to specify the shape of the received var explicitly.

* bugfix for test_cholesky_op

* bugfix for test_compare_op

* bugfix for lookup_table_op

* bugfix for affine_channel_op

* add precision on mac

* added judge

* match file_ut.json on mac

* fix code format error

* fix code format error

* fix error caused by length of ut_lists exceeds the limit

* fix format error,notest,test=cpu

* fix code format error

* add windows judge on get_pr_ut

* - introduced fix onednn 2.2 version

* - compilation fix

* add precision test for windows ci exclude check_added_ut

* fix error

* added PRECISION_TEST parameters

* fix format error

[Kunlun]Multi xpu dygraph performance optimization , add distributed.spawn support for multi xpu and some bug-fixes (#31130)

prepare remove grad op and kernel script.
update Paddle_CI_Inference pipeline.

* [ROCM] update fluid elementwise op for rocm (part10), test=develop

* update, test=develop

* address review comments, test=develop

* [ROCM] update fluid operators for rocm (part3), test=develop

* fix clang format error, test=develop