* ps gpu dump

* remove log

* merge dense

* log level

* tensor copy sync

* format

* pscore support heterps

* fleet cmake

* fleet wrapper

* macro

* solve conflict

* solve conflict

* add unitest

* paddle enforce

* unitest

* unitest

* unitest

* topo and memory performance for heterps; test=develop
* add trainwithprofiler in heter trainier; test=develop

* build gpu task core

* format

* 　add heter box

* add trainer, worker, wrapper...

* format

* for ci

* format

* remove boost get

* boost & copyright

* rename

* 　rename

* format

* format

* format
Co-authored-by: Nyaoxuefeng6 <yaoxuefeng@baidu.com>