Add CLIP to the model zoo

[Rusteam]

What changes are proposed in this pull request?

Adds:

• a CLIP ViT-Base-32 model to the model zoo. It's available for classification and embeddings.
• zero-shot model zoo tag to let users pass custom class labels to classification. Usage template is updated to include an example of passing custom class labels. (by default it will use voc_labels.txt)
• Simple text tokenizer required for CLIP's text encoder.

How is this patch tested? If it is not, please explain why.

Added an extra unit-test for zero-shot label predictions.

Release Notes

Is this a user-facing change that should be mentioned in the release notes?

• No. You can skip the rest of this section.
• Yes. Give a description of this change to be included in the release
  notes for FiftyOne users.

Make a zero-shot class label predictions using CLIP model by using default labels or passing your own labels.

What areas of FiftyOne does this PR affect?

• App: FiftyOne application changes
• Build: Build and test infrastructure changes
• Core: Core fiftyone Python library changes
• Documentation: FiftyOne documentation changes
• Other

[brimoor]

Awesome work, thanks for the contribution! 💯 🥇

Just left a couple formatting requests for you

[Rusteam]

Awesome work, thanks for the contribution! 💯 🥇

Just left a couple formatting requests for you

Thanks it was a bit challenging. I guess because it's intertwined with the eta package and the bulk of the code is based upon config parsers. Overall, it was an interesting experience.

Hope it will be useful for the community.

[Rusteam]

Hi there. Can we get this one merged? I can rebase this branch to v0.16.5

[brimoor]

Hey @Rusteam absolutely, merging now! This will be included in fiftyone==0.17 being released later today 🎉

[brimoor]

LGTM!

[Rusteam]

Thanks a lot. Just preparing to use it actually.