Merge branch 'tfl/gpu_add_nvrtc' of github.com:NervanaSystems/private-ngraph-cpp into tfl/gpu_add_nvrtc

* initial support for hybrid transformer

* add broadcast_vector_rowwise_reversed for hybrid test

* headerc

* get function placement fix

* conv ref test generator graph node in labmda fuction

* rename map_parameter_to_source_node

* type change map_parameter_to_source_node

* use interpreter for numerical derivative

* better comments

MKLDNN optimized layouts: convolution bprop, avg_pool fprop and bprop

Add simple method to add default values for missing deserialized fields. Provided to ease backwards compatibility. (#547)

* working

* cleanup

* fix maxpool copy_with_new_args

* fix a free(nullptr) error

* update output shape from a maxpool

* remove free gaurd

* added Dockerfile.ngraph_cpp.centos74_cmake3, run_as_centos_user.sh, and run_as_ubuntu_user.sh to enable build/test with centos74

* added RUN_AS_USER variable as input for contrib/docker/Makefile

* moved dockerfile for centos7 to contrib/docker

* updated Makefile to take OS as a parameter from the command line for contrib/docker make targets

* fixed capability for setting OS variable on the command line for the contrib/docker/Makefile

* setting the NGRAPH_USE_PREBUILT_LLVM only if the gcc version is not 4.8

* setting NGRAPH_USE_PREBUILT_LLVM for all ubuntu builds

* updated contrib/docker/Makefile to enable the OS=centos74 override for make check_gcc

* More op doc

* Add working doxygenclass import to reST file

* Cleanup punctuation on wording in tables, no punctuation for fragments, remove autodoc extension since we use breathe

* Support Github rendering of cc-README

* Fix sp for basic accessors and update to show mathcal syntax in rst

* modify earlier change after checking what is supposed to render

* Resolve spacing error on rst syntax

* updates for PR to be even with master

* fix numeric stability bug in autodiff of divide

* add a test for divide autodiff stability

* Basic assignment pass for CPU backend

* Change CPU convolution emitter to check op annotations

* Queries MKLDNN for optimal layout on convolutions

* Added layout conversions through ConverLayout ops and explicit layout conversion on CPU tensor view objects

* Added layout conversions for non-MKLDNN ops

* - Style fixes
- Removed unused variables to avoid clang errors
- Added more mkldnn format types to utility functions

* Move ConvertLayout back to runtime::cpu::op namespace

* Added more mkldnn memory formats

* Moved op annotations to Op class

* Style changes

* Minor fix

* Minor fix to keep clang happy

* Use ngraph element type instead of c_type_string in MKLDNN utility functions

* Addressed PR(#502) comments