Since sharing the async runtime across tests makes it hard to debug, and sharing a client across runtimes will cause issues, we should update our tests so that each one creates a separate client.