Add HTTP timeout config for PutRecords call

[jpaskhay]

We're running into an intermittent issue in our environment where the PutRecords call in sendCurrentBatch hangs in the AWS SDK and times out after 5m30s (we see a read timeout when we patched in appropriate LogLevelType settings into the aws.Config). When this happens, the request is actually consumed on the Kinesis side, but there appears to be a potential bug in the AWS Go SDK that causes it to fail to parse the response and then retry internally (resulting in duplicates).

We'll be following up in the AWS Go SDK repo in parallel, as this seems similar to issues that have been previously reported for other Kinesis API calls.

In the meantime, we'd like to have the ability to use a smaller timeout to at least reduce the time spent hanging when this occurs. I'll be submitting a PR that allows for configuring the HTTP timeout.

[jpaskhay]

I forgot to update this issue to link to aws/aws-sdk-go#4149 that was mentioned in the original description.

[zhonghui12]

This change will be included in our next release.