Add a new quickstart example TabNet

[m-danilow]

Reference Issues/PRs

What does this implement/fix? Explain your changes.

Flower has many great examples for various frameworks/architectures, etc., but lacks examples for tabular data which is very popular. TabNet provides a high-performance and interpretable tabular data deep learning architecture. The example utilizes the iris dataset for classification.

Any other comments?

Base of example comes from -> repo
Details about TabNet -> paper

As this is my first PR in this repo feedback will be appreciated :)

[tanertopal]

@m-danilow really cool example! Thank you for it! On my first quick glance, I started making suggestions but then thought maybe you could run poetry run python -m black -q examples from the repo root directory once so it's autoformatted in the default style.

[m-danilow]

@tanertopal thanks, I didn't know black package before :)

[tanertopal]

@tanertopal thanks, I didn't know black package before :)

Yea, it resolves all discussions about formatting. :)

[tanertopal]

@m-danilow The PR looks good to me! @charlesbvll is going to have a look as well tomorrow. Afterwards, we can merge it! Thanks so much! I would be excited to have a short call to understand your background. Let me know if you would be in for it here or on Slack.

Short sidenote: We have to fix a dependency issue related to tensorflow-cpu, which might slightly delay merging your example.