Provide guidance regarding resource requests/limits in EKS

[rapphil]

The defaults for EKS running on EC2 and Fargate are different. If you don't specify resource requests/limits in a pod with EKS for Fargate, the default is to provide 0.25vCPU and 512MB of memory for a node. This might lead the collector to fail to fail to publish metrics in a timelly manner.

What I expect to consider this task done:

• In the EKS sections of our documentation, point out that specifying the limits for Fargate are a good practice, otherwise the defaults of 0.25vCPU and 512MiB will apply.
• Add section to help customers understand when the collector is failing to achieve the performance objectives.