k8sensor needs to support multiple backends

When using Instana, most of our customers require high availability.  They assume that our SaaS operations are highly available based on our SLAs.   But, when customers deploy self-hosted environments they need to set things up in an HA configuration.   While it is technically possible to setup a kubernetes environment with multiple zones and rely on kubernetes to replicate, it is practically impossible in most customer environments due to latency requirements.  In addition, docker and our new Standard Edition don't have HA configurations.

The only way to achieve HA is for the Instana Agents and sensors to be configured with multiple back-ends.   For most sensors, that's no problem because the sensor sends its data to the local Agent.   But, the k8sensor is different.  It sends data directly to the Instana servesr.   Therefore, it needs a configuration option to support multiple back-ends.    

Today, it is possible to set it up with multiple instances of the k8sensor running.  But, that is VERY inefficient.  It means that two different sensor instances are gathering ALL of the kubernetes performance and configuration data for the entire cluster.  Customers want solutions with low overhead.  

Finally, it would be much better if the k8sensor picked up its configuration from the Host Agent.   I configure the Host Agent to use multiple backends.  I can do this via the helm install or by editing the instana-agent configmap.   But, the k8sensor does not pickup and use this additional configuration.   It makes no sense for a customer to configure this information twice.