* [MLU] add fleet init api and collective api pytest for mlu

* fix no value for argument 'data_type' in method call

* fix benchmark and communicator config

* fix bugs of the_one_ps

* multi program and fix bug in optimizer

* multi program in the_one_ps

* public commcontext

* ps optimizer multi programs

* the one ps merge

* fix bug in test

* update

* fix bugs

* modify code style

* fix bugs of _get_global_group

Co-authored-by: NChen Long <1300851984@qq.com>

* Added HCCL backend support in dynamic graph mode

* fix segmentation fault

* add ut

* add recompute for pp

* add recompute offload

* add recompute partition

* fix the docs bug in collective file

Fix the bug in paddle.distributed.split demo. The paddle.distributed.split op just can be used in static mode. (#34306)

* fix the bug in paddle.distributed.split demo

* fix the bug that happened in alltoall in dygraph mode

* cache core.ops, test=develop

* refine, test=develop

* opt dygraph python code, test=develop

* refine, test=develop

* add pplayer

* add sharedlayerdesc

* add parallel_cross_entropy

* add grad for crossentropy

* fix cross entropy

* fix c_split bug

* fix utest

* add c_embedding for tensorparallel

* fix hang of hybrid parallel

* fix new_group for hang problem

* update mp

* fix precision of mp

* fix bug of seed

* fix dp

* print group

* add alltoall api, test=develop

* fixed docs.

* Fixed docs. test=document_fix

code bak.

fixed docs. test=document_fix

* Revert to previous version of python/paddle/fluid/backward.py

* fixed bugs.

* test=document_fix. Fixed examples.

* Support list and tuple for parameters of layer_norm, multiprocess_reader, DatasetFolder and ImageFolder.

* add unittest for layer_norm.

* add require gpu for example.

* add sendrecv, test=develop

* fix model parallel

* rm parallel_help.py

* add embedding

* update