Missing cluster argument from feeding methods

[sumpfork]

During a cloud deployment I'd like to call app.feed_data_point() on the just deployed Vespa app, but it does not take a cluster argument and arbitrarily can call the wrong cluster's endpoint on our multi-cluster deployment.

The cloud deployment's get_endpoint() method does take this argument, so hacking something like

    app.end_point = vespa_cloud.get_endpoint(
        auth_method="mtls",
        instance="foo",
        environment="dev",
        cluster="bar",
    )

before trying to feed works, but it seems like the feeding methods should take a cluster argument and pass it through to determine which endpoint to use? Or else, the app should be initialized to point to a specific cluster?