Spark JobServer should bundle AWS IO libraries to support staging on S3

The AWS filesystem libraries are not included in the Spark JobServer. These are necessary to allow staging artifacts to S3.

There is a TODO in the code to update the dependencies to support the AWS filesystem.

https://github.com/apache/beam/blob/master/runners/spark/job-server/spark_job_server.gradle#L62

Additionally, the `amazon-web-services2` library is built without a dependency on STS. A dependency on STS is needed to support authentication from EKS with IAM roles for service accounts.