* hidden the explicit setting of dataset for pipeline training.

* add device_worker for pipeline, test=develop

* modify pipeline optimizer, 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 a single_process_multi_thread transpiler.
* Add some UTs.
* Fix some API description.

move the thread-check code from train_from_dataset to a single function
add UT for the thread check function