Account for rate limiting in AWS fetchers

[orouz]

Motivation
we need to account for rate limiting in our AWS fetchers to avoid losing resources we want to evaluate.

Definition of done

• figure out the quotas for each AWS fetcher
• when applicable, AWS fetchers method usage does not exceed the default quota
• add a retry (with backoff) mechanism for failed requests

Out of scope

• synced cloudbeat instances consuming the same quota

[orouz]

turns out AWS Go SDK v2 comes with a built-in retry mechanism. verified by modifying the aws clients config to include logging:

awsConfig.Logger = logging.NewStandardLogger(os.Stdout)
awsConfig.ClientLogMode = aws.LogRetries | aws.LogRequestWithBody | aws.LogResponseWithBody 

the important log we get to see is:

Amz-Sdk-Request: attempt=1; max=3

which corresponds to the default settings of the Standard retry of 3 attempts.

regarding quotas for AWS APIs, there doesn't seem to be any publicly defined general limits nor does the client tries to take that into account, and looks like retrying-with-backoff is the approach taken. (for List/Describe/Get requests)

in terms of modifications to cloudbeat, i think adding 429 HTTP status code to Retryable Errors would be sufficient to ensure we retry when APIs are rate limited (this is a safe-measure addition, as a similar retry condition is already present)

adding the additional retry condition, should also be done with consolidation of AWS config initialization as we're currently doing it in 2 ways:

1. using the config provider, which uses libbeat
2. directly using libbeat