[MKLDNN] fix uint8 batch norm memory misuse

[ElaineBao]

Description

When using MKLDNN batch norm, uint8 input needs to be reordered (since MKLDNN bn only supports int8 input). This pr is to address the misuse of original data memory and reordered data memory.

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

• Changes are complete (i.e. I finished coding on this PR)
• All changes have test coverage:
• Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
• Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
• Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
• Code is well-documented:
• For user-facing API changes, API doc string has been updated.
• For new C++ functions in header files, their functionalities and arguments are documented.
• For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
• Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
• To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Changes

• fix uint8 bn reorder memory
• update resnet152-v2 int8 acc

Performance

model	FP32 (Top-1 / Top-5)	Fusion + int8 bn
resnet152-v2	76.65% / 93.07%	76.25% / 92.89%

@pengzhao-intel @ZhennanQin @ciyongch

[pengzhao-intel]

@ZhennanQin please help take a review.

[ZhennanQin]

Please add a uint8 test for this.

[ElaineBao]

Please add a uint8 test for this.

uint8 bn test case has been added at:
https://github.com/apache/incubator-mxnet/blob/master/tests/python/quantization/test_quantization.py#L675-L678

[ZhennanQin]

OK. LGTM.

[pengzhao-intel]

Thanks for the fixing and merging now.