[ECR] [request]: custom S3 bucket for ECR repositories

**Tell us about your request**
As of now ECR is using S3 to store the images. The specific S3 buckets are managed by AWS. It could be interesting for an user to configure a custom S3 bucket per repositoriy or per account (which would work for all repositories in that account).

**Which service(s) is this request for?**
ECR

**Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?**
We have some interests in copying some images from one account to another instead of managing the permissions for the repositories. Other users are also asking to be able to replicate images between regions (see : https://github.com/aws/containers-roadmap/issues/140).

If we were to be able to configure the S3 buckets used by the repositories we could use native S3 features such as same-region/cross-region replication.

**Are you currently working around this issue?**
We are currently working around the issue by fetching the images from one ECR repo, store them as a tarball into S3 then doing to opposite from another account (fetch from S3, push to ECR). Obviously this process is slow and expensive.

No files, tests, or implementation entry points are identified. Start by reading the ECR request and the linked cross-region replication issue 140, then clarify whether custom S3 storage is supported per repository, per account, or both; done means the chosen configuration enables the requested image replication workflow.

Written by the indexing model from the issue text.