fix: workaround to support multiple kubernetes clusters

[rophy]

See #170 for the issue description.

This PR leverages the fact that the plugin actually works for multiple kubernetes clusters if kubernetes:config_file is avaiable in node attribute.

If config_file does not exist and url + token exists, plugin now generates a config_file in project var tmp dir, and sets that in node attribute.

Tested using kind:

$ env | grep RD_ | sort
RD_CONFIG_ATTRIBUTES=
RD_CONFIG_NAMESPACE=kube-system
RD_CONFIG_TAGS=tag.selector=default:status
RD_CONFIG_TOKEN=***
RD_CONFIG_URL=https://127.0.0.1:38287
RD_PLUGIN_VARDIR=/tmp

$ python contents/pods-resource-model.py 
INFO: kubernetes-model-source: ------
INFO: kubernetes-model-source: container-name:coredns
INFO: kubernetes-model-source: pod-container-name:coredns
INFO: kubernetes-model-source: Generated /tmp/tmp/a59e74c1b8100831
INFO: kubernetes-model-source: ------
INFO: kubernetes-model-source: container-name:coredns
INFO: kubernetes-model-source: pod-container-name:coredns
...
[
    {
        "default:container_id": "containerd://069c6a2413c36c466d9023bf3b9853524d675b66f8af55d248b289d8367caa42",
        "default:container_name": "coredns",
        ...
        "kubernetes:config_file": "/tmp/tmp/a59e74c1b8100831",
        "labels:k8s-app": "kube-dns",
        "labels:pod-template-hash": "6d4b75cb6d",
        "nodename": "coredns-6d4b75cb6d-2trxl-coredns",
        "status": "running",
        "tags": "pods,running",
        "terminated": false
    },
    ...
]