Default to standard retry

[mullermp]

Sets the default retry strategy to standard.

This is a mild breaking change, rendering some config values useless (since standard doesn't have too many configurable options).

DynamoDB extended retries will include a new default of 10 retries but no configurable jitter/backoff policy unless configured back to legacy.

Other changes to the retry plugin are rubocop related.

[github-actions]

Detected 7 possible performance regressions:

• aws-sdk-dynamodb.get_item_small_ms - z-score regression: 0.59 -> 0.69. Z-score: 18.5
• aws-sdk-dynamodb.put_item_large_ms - z-score regression: 58.07 -> 73.48. Z-score: 33.79
• aws-sdk-dynamodb.batch_get_item_large_ms - z-score regression: 76.76 -> 88.85. Z-score: 35.43
• aws-sdk-kinesis.put_record_small_ms - z-score regression: 0.44 -> 0.51. Z-score: 15.14
• aws-sdk-s3.get_object_large_ms - z-score regression: 3.73 -> 6.82. Z-score: 18.79
• aws-sdk-s3.put_object_multipart_large_ms - z-score regression: 33.6 -> 39.97. Z-score: 22.86
• aws-sdk-sns.publish_small_ms - z-score regression: 0.53 -> 0.62. Z-score: 16.27

[jterapin]

LGTM ✨

[jterapin]

I think it would be worthwhile to link this resource: https://docs.aws.amazon.com/sdkref/latest/guide/feature-retry-behavior.html

I believe this documentation is very current since it was linked to the blogpost as a point of reference: https://aws.amazon.com/blogs/developer/updating-aws-sdk-defaults-aws-sts-service-endpoint-and-retry-strategy/

[jterapin]

I was expecting see this configuration to be mentioned in the AWS SDK documentation on retries. If this is necessary, we should probably see if someone could update the documentation.

[jterapin]

Was there a reason why it was 10 attempts? I was curious

[mullermp]

I'm not certain. Probably because of throttling and can be loss of data.