Affine transformation APIs for high-performance image augmentation

[zsdonghao]

Checklist

• I've tested that my changes are compatible with the latest version of Tensorflow.
• I've read the Contribution Guidelines
• I've updated the documentation if necessary.

Motivation and Context

Support Fast Affine Transformation

[zsdonghao]

@luomai please write the docs, and also the dataset prefetch order of existing examples.

• docs/modules/prepro.rst
• examples/data_process/README.md
• examples/basic_tutorial/tutorial_cifar10_datasetapi.py
• check https://github.com/tensorlayer/tensorlayer/pull/855/files

[DEKHTIARJonathan]

Please do not merge, explanation below.

After some discussion with @zsdonghao, we stated that only bug fixes and dependency updates will be merged. This is done to ease transition toward TL 2.0. We need to keep an easy and clean codebase with the change upcoming both with TL 2.0 and TF 2.0. Every-time big changes are performed, additional work is required to reflect the changes in the PR #755, in particular if those changes impact the layers module.

You can still update this PR as your wishes. It will be merged as soon as possible once we have cleared the backlog.

If you need this PR to work, you can manually build this PR and install it locally ;) Sorry for the inconvenience.

Thanks for your understanding and the work ;) We'll get back to you as soon as we are ready to merge.

[luomai]

@DEKHTIARJonathan Hey John, this PR would not touch any layer. It only affect tl.prepro.

[zsdonghao]

@DEKHTIARJonathan this PR would not affect the layer APIs, and we really need it for the openpose project.

Merge possible without work

[DEKHTIARJonathan]

@zsdonghao @lgarithm if you want to want to merge this PR, LGTM.
Thanks to @luomai, the merge commit it already done in TL 2.0

Before merging please fix these issues: https://app.codacy.com/app/zsdonghao/tensorlayer/pullRequest?prid=2273850

fixing codacy