Update remaining DockerHub-hosted repositories to use ECR

[007]

tl;dr: Please update amazon/foo registry values in these charts to use Amazon-hosted ECR versions of the same.

Per the Coming Soon blogpost for the AWS public registry:

All EKS add-on software that is included with EKS clusters is hosted on ECR today and will not be subject to Docker Hub rate limits. EKS-built and maintained projects, such as the AWS Load Balancer Controller include helm charts that reference images on ECR as well. However, manifests for these projects available on GitHub include references to images that are published to Docker Hub. You can check whether you are using a manifest with a Docker Hub image using the steps described in this blog and update to use a chart that references an ECR hosted image. These charts are available in the EKS Charts GitHub repository.

We went to update our aws-node-termination-handler to use this chart after getting DockerHub pull failures, but it doesn't look like it's going to make a difference:

eks-charts/stable/aws-node-termination-handler/values.yaml

Line 6 in d721c97

repository: amazon/aws-node-termination-handler

There are several other places that would be subject to the same throttling:

stable/appmesh-controller/values.yaml:45:    repository: amazon/aws-xray-daemon
stable/aws-cloudwatch-metrics/values.yaml:2:  repository: amazon/cloudwatch-agent
stable/aws-for-fluent-bit/values.yaml:6:  repository: amazon/aws-for-fluent-bit
stable/aws-node-termination-handler/values.yaml:6:  repository: amazon/aws-node-termination-handler

And several public images that might need to be replicated as well, either as part of the deployment or as part of CI:

stable/appmesh-controller/ci/values.yaml:7:  repository: fawadkhaliq/appmesh-controller
stable/appmesh-gateway/ci/values.yaml:5:  repository: envoyproxy/envoy
stable/appmesh-grafana/values.yaml:8:  repository: grafana/grafana
stable/appmesh-inject/ci/values.yaml:5:  repository: stefanprodan/aws-app-mesh-inject
stable/appmesh-jaeger/values.yaml:8:  repository: jaegertracing/all-in-one
stable/appmesh-prometheus/values.yaml:11:  repository: prom/prometheus
stable/aws-load-balancer-controller/ci/values.yaml:5:  repository: kishorj/aws-load-balancer-controller

Please update the remaining amazon/* images (at least) to use an AWS ECR repository value instead of relying on the current DockerHub repo.

[bwagner5]

aws-node-termination-handler will be moving to ECR images soon.

[fawadkhaliq]

AWS ECR public gallery is out: https://gallery.ecr.aws/

We'll be moving the images to ECR gallery

[bwagner5]

NTH Public ECR: https://gallery.ecr.aws/r6b0f9a1/aws-node-termination-handler

Chart will be updated soon to use it by default. Images will still be pushed to Dockerhub in addition to ECR for now.

[mamoit]

The cloudwatch-agent's image tag that comes by default in the chart is not available in the public repository.
It seems to be too old.

public.ecr.aws/cloudwatch-agent/cloudwatch-agent:1.247345.36b249270

Should the default image tag be bumped?
And maybe even setting the default registry to be the public one to not depend on dockerhub.