Correctly check inference endpoint is ready

[Wauplin]

From @McPatate on slack (internal):

The condition in the InferenceEndpoints::wait function isn't what I expect to mean "ready":

        while True:
            self.fetch()
            if self.url is not None:  # Means the endpoint is deployed
                logger.info("Inference Endpoint is ready to be used.")
                return self

as the url can be provisioned before the endpoint started. Wasn't it that we made it so that huggingface_hub queries the url until status is 200?

This PR fixes this by making a GET request on the returned URL before returning (in .wait call).

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[McPatate]

lgtm!

[Wauplin]

Thanks for the review :)