kconnect throttling describeCluster AWS API

[ashishapy]

Describe the solution you'd like:
kconnect to optimize for high volume calls. kconnect to optimize how it makes calls to AWS API.

We made 100 concurrent call using kconnect & approx 20 calls failed with below error

Failed to connect to cluster Error: getting cluster: getting cluster config for arn:aws:eks:<account$>:cluster:<cluster-name>: describing cluster <cluster-name>: TooManyRequestException: Too Many Requests

Why do you want this feature:
For scalability & stability

Anything else you would like to add:

AWS Support has shared below documents to implement retries & exponential backoff. AWS confirmed that describeCluster is being invoked while generating KubeConfig from EKS API.

https://docs.aws.amazon.com/sdkref/latest/guide/feature-retry-behavior.html
https://aws.amazon.com/blogs/architecture/exponential-backoff-and-jitter/
https://aws.amazon.com/builders-library/timeouts-retries-and-backoff-with-jitter/

Proposals:

• Implement retries & exponential backoff as per AWS recommendations.
• Provide a way to generate static kubeConfig for each cluster only once & regenerate only sts token when trying to reconnect to same cluster using same user or different user.

[antisec135]

We can add a sleep functionality if AWS API throws TooManyRequestException error

[antisec135]

Introduced MaxRetries to #770