* remove SliceCopied

* remove SliceCopied

* rename SliceShared to SliceLevels, SliceInLevel

* merge lod_tensor/details

* remove lod_start_pos_'s shared_ptr

* make lod-tensor a special tensor

* add clone to lod_tensor

* add lod tensor impl

* add lodtensor clone test

* init lod

* add equal

* merge LOD and its methods

* recover tensor and variable

* change thrust to host_vector

* golang pserver optimizer config for user

* update

* update

* update

* update

* update by comments

* fix errors

* fix errors

Otherwise the production image generation will fail due to deb not
found, or executable built from Go not found.

related codes in Paddle core.

to create a gradient machine for inference using merged model with parameters
which is genearted by `paddle merge_model`.

* Should always use `dynload::` for cuda function.
* Fix cublas.h without DSO load.

Fix #3386