(slating for 8.2.0 as it will be potentially useful for Kangaroo)

New storage option was implemented (but not tested in real life) and documentation was added to the project.

Build/documentation fixed.

@andrzej.wojcik after quickly taking a look - I think it would be prudent to include default-virtual-host in the key. In theory we could use it for bucket name, but with many installations this would quickly pollute the list of buckets. Having something under one bucket and differentiate installations by main-vhost-keys seems quite useful - what do you think?

@wojtek Being able to set a prefix for a data stored within a bucket could be useful. However, I do not agree that we should use main-vhost-keys as this most likely will be problematic (someone wants to change vhost?, etc.)

I suppose it would be better to allow to set (in the config file) bucket-prefix which, if present, would be added as a prefix to the file key in the bucket. (this prefix could be set in the EC2 instance enviroment and retrieved by TDLS.

+1

Prefix wise, it could be configure to use same environment variable as main-vhost-key.

I've added bucketKeyPrefix property which when not null and not blank will be used as a prefix for S3 key generation for uploaded/downloaded files. I've updated documentation as well.

Moved documentation to http-api project with appropriate NOTE about requirement of extras binaries and dependencies. Included all required dependencies in -dist-max distribution packages.

I was pondering updating to aws sdk2 to limit number of dependencies (it currently allows using internal JVM http client instead of using Apache's: https://aws.amazon.com/blogs/developer/aws-sdk-for-java-2-x-released/, https://docs.aws.amazon.com/sdk-for-java/latest/developer-guide/setup-project-maven.html#sdk-as-dependency) but the potential saving would be minimal.