High Availability (HA) for Instana AWS Agent (Cloud Watch Kinesis Stream)

Introduction:
-------------
We want to make heavy use of the AWS Instana Agent on various AWS Accounts 50+. The Teams want to see and use the AWS Metric Data in Instana for Alerting and Analysis.
Therefore we plan to set up an Instana AWS Agent. We learned that we could only use one dedicated instance of the AWS Agent per Account from the past and other customers. This means a failover of the Instance will cause metric data loss, which we want to prevent or keep as small as possible.
We plan to set up a 1:1 AWS Autoscaling group, which setups everything automatically. This will also have a turnaround of min. 5 min, which feels too long for us.
Therefore we ask for a better solution, enabling us to use Kubernetes or AWS Instances with multiple machines (warm standby) with the leader elector principle.

Use-Case:
----------
- When an Instana AWS Agent or its underlying hardware crashes, a warm standby AWS Instana Agent is immediately started, with almost no metric data loss.
- When an update is executed on the agents, the AWS Agent will hand over the metric collection to another Agent without metric data loss.
- We can use the AWS Agent on Kubernetes / Openshift like we use a Kubernetes Leader Agent.

Desired Solution:
-----------------
Add a Kubernetes sensor like Leader / Elector principle for Instana AWS Agents on an Account basis. This should enable us always to have exactly one Instana AWS Agent per AWS Account active. Of Course, an Instana AWS Agent should be able to manage multiple AWS Accounts.

Ticket: https://instana.zendesk.com/agent/tickets/17292