* "add c++ side kernel selection"

* "add multiple kernel op test"

* "kernel selection only support cudnn"

* "better formatter"

* "small fix with UseCPU"

* "depends on change interface Get(Place, Library)"

* "fix CI"

* "fix python cudnn test"

* "leave the register cudnn op to another PR"

* "fix CI"

* "use all kernel by default"

* "fix CI"

* send_recv variable

* delete unused logs

* fix ci failed

* update

* resize tensor before tensor copy

* add selectedrows unit test

* check rows

* "remove cudnndevicecontext"

* "remove unused init code"

* "fix hash functions"

* "fix typo"

* "fix based on comments"

* "follow gogle style"

* "fix based on comemnts"

* fix bug in SetAttrDescVisitor

* add comments

* "fix data transform"

* "data transformer"

* "add device pool"

* "add test"

* "fix ci"

* "fix datalayout implementation "

* "fix based on comment"

* add helper function to get appropriate DeviceContext

* Run interface return future

* delete unused comments

* implement selectedrows serialize and deserialize

* make serialize/deserialize as global function

* recover send_imp.cc

* delete unused brackets

* fix compile error

* serialize version in LodTensor and SelecetedRows

* fix ci

* fix ci

* Rename API of DeviceContext

Make them as usual names.

* Rename API of DeviceContext

Make them as usual names.

* Fix compile

* Fix compile

* Fix compile

* Fix compile

* Fix compile

* "fix data transform"

* "split into next PR"

* fix bugs

* add KernelTypeToString interface

* cache memory in local scope

* fix typo

* refine trans logic

* "refine kernel registrar"

* "refine registrar with multikey"

* "fix register"

* "refine multikernel register"

* "fix CI"

* "fix CI"

* "fix registry"

* "swtich GPU to CUDA"

* "add register macro test case"

* "fix CI"