fix: update eks in-cluster authentication to use web identify files

[mrsimonemms]

Description

The previous #288 installation doesn't seem to work because there's no secret specified in the credentials.NewStaticCredentialsProvider call. I don't fully understand the reasons why this was changed because the envvar AWS_WEB_IDENTITY_TOKEN_FILE is one of the documented ways of authenticating - https://aws.github.io/aws-sdk-go-v2/docs/configuring-sdk/#specifying-credentials

This meant that the /regions/aws call always returns the error failed to refresh cached credentials, static credentials are empty.

@jarededwards I'd like to know if there is something I've misinterpreted in here. I also can't find any other use of the aws.NewEKSServiceAccountClientV1() call in the API, so I think it's safe to use.

Related Issue(s)

Fixes #

How to test

• In your gitops repo, go to /registry/clusters/<cluster-name>/components/kubefirst/console.yaml
• In spec.helm.values, update your kubefirst-api to include the image.api.repository and image.api.tag below

  kubefirst-api:
    image:
      api:
        repository: ttl.sh/90df18fd-62d4-4cdc-8389-05364e3e4c48
        tag: 24h

NB. I've built the kubefirst-api locally and pushed to ttl.sh. This will expire at some point on 2024-07-23 (approx 09:00 UTC) so you may need to rebuild. Do that by running:

IMAGE_NAME=$(uuidgen)
docker build -t ttl.sh/${IMAGE_NAME}:24h .
docker push ttl.sh/${IMAGE_NAME}:24h

You can then update the image.api.repository value in the console.yaml