v0.1.0 A minor milestone

This is Rikai's first minor release. After internal usage for sometime, we'd like to engage the wider open-source community for feedback.
In preparation for this release we've added a number of documentation improvements (though it's still very rough around the edges).

We've also included a major features -- Tensorflow Model Runner, which bring pytorch/tensorflow support in Rikai up to parity. Given the differences between the two frameworks, the user experience with Pytorch models is still easier at this point. We'll look to improve this over time.

Thank you to all Rikai contributors so far:

@darcy-shen
@Renkai
@ffcai
@chunyang
@bobingm
@smellslikeml
@eddyxu
@changhiskhan

What's Changed

• Fix a bug that prevents to use ML_PREDICT in multiple SparkSessions by @eddyxu in #502
• Preliminary support of tensorflow models in SQL ML by @eddyxu in #507
• Ssd by @changhiskhan in #508
• [no ci] Notebook to analyze and visualize the cover of Mojito by @da-tubi in #503
• Add a doc section to describe how to use numpy and tensors by @eddyxu in #510
• Improve default GPU behavior in Torch runner by @eddyxu in #512
• Set minimum ipython for CVE-2022-21699 by @da-tubi in #511
• Fix rtd build with missing dependency by @eddyxu in #514

Full Changelog: v0.0.21...v0.1.0