Submit Competition Code: CycleGAN Related

[bravotty]

In this PR we are submitting our code implementation for the competition.

In general, we implement an improved CycleGAN with TinyMS & MindSpore. Specifically, we guide the image translation task between different domains by constructing new regularizier.

Please notice the code implementation we submited here does not represent the final version. In the future we will make more improvements based on it.

[CLAassistant]

All committers have signed the CLA.

[hellowaywewe]

@bravotty Thank you for your submission! Please sign our Contributor License Agreement before we can accept your contribution.

[bravotty]

@bravotty Thank you for your submission! Please sign our Contributor License Agreement before we can accept your contribution.

Thank you for reminding me. I have signed the CLA.

[hellowaywewe]

Hi, @bravotty It is better to add supported hardware platforms to this file

[hellowaywewe]

Maybe you need to be familiar with the tinyms directory architecture, put your core code into the tinyms/tinyms module directory, and put the main.py and other executable code into the tinyms/tests/st directory, create a new directory named competition under this directory, then place the city_cyclegan.py execution code, which can avoid repetitive code and help other developers use it quickly.