add data_generator into paddle.fluid.incubate.data_generator, add op run log in hogwild_device_worker and downpour_device_worker
test=develop

test=develop

dataset_generator.py is a framework for generating data with python
the generated data with a fixed format will be feeded into c++ reader
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

* fuse optimizer

* Make unitest of fsp op faster and more stable.
test=develop

* Skip unitest of fsp op.
test=develop

test=develop