graphql requests module

[horheynm]

Use v2 api endpoints for search and downloads.

Please checkout the graphiql interactive UI to play around with the apis.
https://staging-api.neuralmagic.com/v2/graphql

Example query:

    {
        models (domain: "cv",task: "classification",architecture: "resnet_v1",subArchitecture: "50",framework: "pytorch",repo: "sparseml",sourceDataset: "imagenet",sparseTag: "pruned95_uniform_quant-none",)
            {
               modelId modelOnnxSizeCompressedBytes stub 
            }
    }

[dbogunowicz]

Left some tiny comments.
In general, as a reviewer I would love to have this split to one big feature branch and smaller PRs (much easier to review). Also, wouldn't hurt to leave more verbose docstrings.

[horheynm]

Left some tiny comments.

In general, as a reviewer I would love to have this split to one big feature branch and smaller PRs (much easier to review). Also, wouldn't hurt to leave more verbose docstrings.

Thanks for the comments. A good reason for the length of the pr:
Ideally wanted to make prs for both search and downloads. But when search is called, downloads gets called in the original main branch's code (without downloading the actual file, just gets the download url). Was not easy to decouple the logic without making larger changes (will need to edit the Model class, and completely decouple load_files_from_stub)

[tdg5]

Nothing glaring sticks out to me, though I would be interested to know if there's a good reason for making multiple graphql requests for a single model

[corey-nm]

Looks great! Made a couple suggestions on structure, most are nits though. Left one comment about the sparse_tag, that's the only thing I saw that looked potentially wrong

[corey-nm]

Nice, looks good to me! 👍

[dfneuralmagic]

LGTM