* rename sendrecv.proto to namespace paddle.distributed

* split ps with distributed

* oneps (3/4)
Co-authored-by: NMrChengmo <cmchengmo@163.com>
Co-authored-by: Nmalin10 <malin10@baidu.com>
Co-authored-by: Nchengmo <chengmo@baidu.com>

* add heter ps mode

* code style
test=develop

* add with_pslib
test=develop

* unitest
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* code style
test=develop

* test monitor
test=develop

* prepare trainer
test=develop

* code style
test=develop

* fix dump, fix cvm check
test=develop

* fix
test=develop

* fix
test=develop

* fix
test=develop

* Refactor code for dump_field & dump_param: abstracting the common function in base class.
* Support dump randomly & random with lineid
* Support specify the random interval, which avoids printing too much logs.

* add half_async in the communicator
* fix DistributedStrategy

* fix fetch handler problem and refactor
when a user define FetchHandler class, he or she should initialize a handler
with variable dict. the key of a variable dict is a user defined name,
the value of a variable dict is a Varaible generated from python API.

For each fetching, a user should implement handler function in which
fetched_result_dict will be available and the user can access the fetched value
with user defined keys.

add executor.FetchHandler for train/infer from the dataset

* add thread scope stat accurate metrics test=develop

* fix style

* fix style

* fix style

* fix style test=develop

* fix style test=develop

* fix style test=develop

* fix style test=develop

* fix style test=develop

* fix style test=develop

* fix style test=develop

* fix conflict

* fix style

* fix style test=develop

* fix error test=develop

* fix error test=develop

* add a place field in DataFeed to denote which place it will feed data to.
* abstract the copy process in CopyToFeedTensor function
* add UT for float32 type and for CUDAPlace

(1) use channel instead of vector/BlockingQueue in Dataset，to keep same with existing implementation, and make code more readable and flexible (dataset single output channel or multi output channel). one previous memory out of limit problem is cause by not release memory after training.
(2) add Record because MultiSlotType costs too much memory (80B)，fix memory out of limit problem.
(3) add Channel, Archive in paddle/fluid/framework
(4) change dataset from shared_ptr to unique_ptr in pybind
(5) move create/destroy readers from trainer to dataset
(6) move shuffle from datafeed to dataset. dataset holds memory, datafeed is only for load data and feed data to network.
(7) fix thread num bug of Dataset when filelist size < thread num
(8) support set_queue_num in InMemoryDataset

test=develop

test=develop

test=develop

test=develop

add dist_multi_trainer for distributed training, add trainer_factory and device_worker_factory so that we can easily extend new training mode, add pull dense worker which is a singleton for parameter fetching