* fix ports conflict when launching multi-nodes in paddlecloud;test=develop

* add DISTRIBUTED_TRAINER_ENDPOINTS env for cloud;test=develop

* fix strategy, test=develop

* fix can_apply

* rm auto from localsgd

* parameter_server_optimizer support auto_strategy

add lars to fleet meta optimizer

* refine launch and distributed repr string for print

* test=develop, add ps auto

* fix the localsgd

* add auto parallel L1 implementation
test=develop

support heter-xpu-ps

* fix doc

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* fix doc, test=develop

* update localsgd doc
test=develop

* update localsgd doc
test=develop

* fix fleet dgc amp doc, test=develop

* fix, test=develop

* fix async configs
Co-authored-by: Nliuyi05 <gavin1332@gmail.com>
Co-authored-by: NWangXi <wangxi16@baidu.com>
Co-authored-by: NseiriosPlus <tangwei12@baidu.com>

paddle.distributed.fleet supports dynamic graph execution.

* print detailed and clear log infos; test=develop

* support running python train.py for fleet-task; test=develop

* fix fluid.embedding

* add FleetAPI doc
Co-authored-by: Nliuyuhui <liuyuhui@baidu.com>

* Support Heter Parameter Server

* fix api document

* modified timeout value on windows and mac (#26690)

* add Local SGD algorithm referenced paper
test=develop

* add cudnn related strategies to DistributedStrategy

* solve the initial configuration about fleet and rolemaker
Co-authored-by: NseiriosPlus <tangwei12@baidu.com>

* add documentation for DistributedStrategy

* consider the combination of different strategies to work together

* add save/load for parameter server

* add feature to fleet2.0 role_maker, distribute_strategy, test=develop

* add check approval
test=develop

* make fleet_localsgd_meta_optimizer work

* fix bug in localsgd meta optimizer

* set default value to strategy in distributed_optimizer test=develop

* add unittest for paddlerolemaker with gloo

* move paddle.fleet to paddle.distributed.fleet