Automatic requests batching for SQS messages

[bigunyak]

Type: Feature

Describe the solution you'd like
In AWS SDK v1 there was a feature which allowed automatic batching of outgoing SQS messages.
For that we only needed to create AmazonSQSBufferedAsyncClient and supply it to QueueMessagingTemplate. That worked perfectly because AmazonSQSBufferedAsyncClient implements AmazonSQSAsync interface which makes it completely transparent for usage in QueueMessagingTemplate.
Unfortunately, this is no longer the case with AWS SDK v2. New implementation for automatic batching is now based on SqsAsyncBatchManager which makes it impossible to use with spring-cloud-aws. Searching in the spring-cloud-aws codebase, I find no usages of SqsAsyncBatchManager, so looks like it's not supported in any way in the current implementation.
My question is if you're planning to add support for it in the future?

Describe alternatives you've considered
Current alternative is to use SqsAsyncBatchManager directly instead of SqsTemplate.

[tomazfernandes]

Hi @bigunyak, thanks for bringing this interesting new feature from AWS SDK up.

I believe we should support using this client.

Since the requests themselves seem to be the same, we should be able to allow switching to this client based on configuration options both for the SqsTemplate and the SqsMessageListenerContainer.

One thing to be mindful though is that when we use batching we might have false positives - e.g. the template will return successfully but perhaps there will be an issue when the requests are actually being sent asynchronously, so that should be disclosed in the docs.

PRs are welcome, happy to discuss implementation plans.