* Fix batchnorm argument order, cleanup some comments, fix backprop

* Merge error

* Clean up training function, organize inference test

* BatchNormInference tests

* Training case

* Training test

* Fix autodiff BatchNorm test

* Cleanup

* Move file to doc checkout

* Update disabled test name in igpu manifest
Fix unnused variable

* Unit tests disables

* Review comments

This re-enables `CPU.batchnorm_bprop_n4c3h2w2` was disabled. This test seems to have a checkered past:

* Back when it was called `cpu_fusion.bn_bprop_n4c3h2w2` we were observing some random crashes on macOS; this was before we officially supported macOS, so we didn't worry about it.
* More recently, the test was disabled in #2058 because it "just started failing" (no further details available).

I have tried to reproduce the failure with repeated gtest runs as follows, with no success:

* macOS Mojave with DEX
* Ubuntu 16.04 with gcc-5.4.0 and DEX
* Ubuntu 16.04 with gcc-5.4.0 and codegen (pre-built LLVM)
* Ubuntu 16.04 with clang-3.9 and DEX
* Ubuntu 16.04 with clang-3.9 and codegen (non-pre-built LLVM)

I think the best thing to do is to re-enable it, and fix it if the failure crops up again. (Who knows, maybe CI will trip over it!)

* Fix uninitialized parameter tensor data in nbench

* Remove const modifier in the loop to fix compilation error

* Add condition check before writing tensor data

* Remove the condition check when initializing pramater tensor data

- All "<>" and all \"\" are in two separate groups.

Use @loader_path for the calling library/executable
Use @rpath for dependent libraries.

* disable squeeze layout optimizations for padded layouts

* knob squeeze padded layouts

* remove version tagging from shared libraries

* add flag to enable/disable library versioning

* Update MKLDNN v0.16 -> v0.17

* Updated SHA1 for mac and win

* Updated from 0.17-rc -> 0.17

* Patch to treat OMP as warning instead of error

* Fixed compile error for clang

* Commbined 2 patches into 1 file; renamed file to .patch

* Corrected file permissions

* Adding leaky relu

* Silence compiler warning around fp compares

* Fix copy-paste error and enable in-place for relu mkldnn kernels

* disable in-tree build

* check to prevent in-tree builds

* add cuda reduce for product and max and added tests

* a quick fix for empty reduce axis and 0 axis

* adding min cuda reduce

* add for min

* fix bug and format

* add another min test

* adding sum to the mix and adding tests to intelGPU manifest

* Incorporate Chris's first comment + clang

* Some mods to the last commit

* Addressed Bob's comments + added more tests int8

* Added more int8 tests + added tests to IntelGPU manifest

* CI test failure debug attempt

* clang

* edit

* Adding the CPU failing test to manifest

* pulled changes from master to address travis ci build failure

* Re-enable INTERPRETER tests which already seem to work

* Update INTERPRETER to be aware of TopK's unusual output element type arrangement; re-enable INTERPRETER.topk_int64

* [Py] Add make python_wheel target

* Add make install step

* Use tagged pybind11 version 2.2.4

* build seems to work

* cleanup

* add backsuport for using pre-built clDNN library

* fix typo

* add license badge

* move license badge

* any_of matching

* include a new file any_of.hpp

* hacky version of reshape+convertlayout reordering

* Cleaned up reshape+convertlayout and added unit test

* fix pass-through check

* style fix

* fix ParameterVector

* reduce random input range

* Add more checks in horizontal fusion.

* Address PR feedback: use a helper function.

* more sdl changes

* more fixes

* update SDL

* cleanup

* cleanup

* rename var

* redo how in_transition is handled

* add -O2 flag back

* - check if the nodes is reachable before returning the users of a node

* address PR comments

* test case fixes