* [ONNX] Assert all op types supported

* Apply clang-format

* Address code review comments

* Fix #include statements

* Update ONNX Squeeze Op implementation to conform with doc. Add unit test.

* Apply code-format.

* Correct attribute value type.

* Change used loop structure.

* Modified version of loops.

- Without erase and with minimal computation time complexity.

* Run CI

* Reshape with no permutation optimized out

* PR1671. Apply optimization to the same shapes only

* PR1671. Reuse Reshape optimization from common passes

* PR1671. Remove eliminate_reshape from nop_elimination procedure

Add support for an environment variable `NGRAPH_ONNX_IMPORT_ENABLE` to setup.py
This variable controls whether the Python API is build with or without support for ONNX import functionality built into core nGraph.

* Improved AvgPool unit test coverage. Fixed small bug that was revealed.

* Renamed disabled unit tests to reflect new names.

* Ran clang-format on backend_test.in.cpp to fix format.

* Renamed cpu_results->backend_results in two unit tests.

* Added Jenkins Job No and URL

* Update prepare_environment.sh

- clone onnx models
- remove wheel build artifacts

* Link onnx models

* Update Jenkinsfile

- Only send email on fail,
- Change ONNX branch warning to info field
- Disable cloning after tox tests

* Update prepare_environment.sh

- Added comments
- link ONNX models directory

* Update cleanup

* dynamic to static casts

* ScalarConstantLikeBase needs a dynamic cast

* revert change

*  unary, binary folding

* fix divide wrong template args

* add tests

* fix merge breaks

* Why am I still needing to fix license headers?

* fix a few more in test

* Handle zero sized buffer requests

* Removed aligned buffer initialize() and moved allocation to constructor instead

* onnx: add register operator macro
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: add set information to 'op' namespace
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* initial rough draft

* import changes md to release notes for readability

* import changes md to release notes for readability

* remove draft

* update README with link to Release Notes

* release notes link

* fix graphic placement

* Wording change on ONNX supported frameworks

* return nullptr when workspace size is zero+modify insert method to accept const lvalue ref

* Unit test added

* move backend specific test files to test/backend directory

* remove unit_test_control

* move tests back to test root

* fix comment

* wip

* fix manifest

* update test to verify all header files are complete, meaning they include what they use.

* disable

* Project initialization commit

* Added unit tests for 3D tensors for argmax

* Refactored reduce to be used by argmax argmin. argmax argmin still has some issues. WIP

* [WIP]First working version of ArgMax ArgMin

* added reduce buffer for the cudnn api calls

* added reduce buffer for the cudnn api calls

* Further modifications. Using rvalues to pass enums to build reduce method

* more unit tests added

* Incorporate Fenglei's comments

* Incorporating Chris's first set of comments

* small change to test file

* Resolving clang issue that was causing argmin test to fail

* Incorporate Chris's  comments

* clang format issue

Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* updated unit tests

* remove debug comments

* update onehot

* clang

* fix bugs

* format

* add output_datatype_size to hash

* typo

* hash

* reshape sinking working on mnist_conv

* forgot to add reshape_sinking files

* refactoring of binary case

* Quantize/Dequantize case, fix add case, add assert

* address bob and scott's feedback

* debug

* fix a bug where reshapes are removed too early

* add find algorithm for convolution without extra padding

* Use cudnnFind* or cudnnGet* depending on tuning param boolean. Add select function to search the perf results of the cudnn queries.

* Formatting.

* Algo search no longer binary, now it is either off, a heuristic search (cudnnGet*) or an explicit search (cudnnFind*).

* Formatting.

* switch to explicit.

* Throw if no suitable cudnn algo found.

* Formatting

* Remove comment.

* Added a bounds check for mkldnn layout descriptor creation

* Added dims check

* Add pad with fill operator using the outward-in index pattern.

* Remove static pad and rename build_pad_dynamic -> build_pad. Update maxpool 1d padding.

* Formatting.

* Split build_pad_dynamic into build_pad and build_pad_fill.

* Add test coverage for fixed bug in op::Pad for gpu.

* Reshape optimizations for when unit-sized dimensions are added/removed from tensors

* Added unit tests for eliminating squeeze and expand_dims operations

* Bug fix to expand dims layout

* Style fix

* Check output shape when setting memory layout for slice op.

* Miscellaneous fusion and other optimizations for inception-resnetv2
- ConvBias Batchnorm folding
- ConvBias Affine folding
- Check if MKLDNN can slice a given layout and select layouts
  appropriately

* Fixed unit test and bug in conv bias pattern

* Addressed PR feedback

* Addressed PR feedback

* Eliminated two warnings introduced in #1459

* Removed unnecessary call to reserve_workspace.

* added constant folding for dequantize

* modified as per review comments

* LRN WIP

* Explicit lambda captures.

* Switched to Ayan's new caching routine.

* Remove commented out lrn from manifest.

* Fixed clang 3.9 error.

* Corrected lrn hash. Only call cudnnSetLRNDescriptor once.

* Simplified lrn hash. Removed redundant parameters. No longer passing CUDNN_LRN_CROSS_CHANNEL_DIM1 as parameter because it's the only choice for cudnnLRNCrossChannelForward.