Integrated Trainer of Parameter Server (API add `fluid.contrib.layers.sparse_embedding` only) (#22957)

* Integrated Trainer of Parameter Server

* fix communicator when breaking under PyReader mode, test=develop
* revert some vlog level to 0, test=develop

* add sync communicator and implement

* add half_async in the communicator
* fix DistributedStrategy

* add distributed_strategy

* test=develop, speed dense calc & communication

* remove -Wno-error=sign-compare, make warning as error

test=develop test=document_fix

* fix exist compile warning

test=develop

* update pserver decay blocks

* update distributed notify handler

* test=develop, communicator_is_sgd_optimizer flags

* delete debug vlog & add rpc function & fix word2vec bug & speed GEO-SGD

* refector geo sgd & communicator

* add a base class for the Communicator
* add AsyncCommunicator Impl for async distributed training

* test=develop,  communicator merge add => merge average

* fix correctness of the communicator

* fix a bug in send thread when sending var context is empty, test=develop

* add lookup_table_prefetch_op and prefetch optimize, test=develop

* remove remote prefetch GPU supported

* word2vec force with CPU, test=develop

* test dist remote lookup table force with CPU, test=develop

* add is_runnning in communicator, test=develop

Async exe support communicator