correct and clean manifests and cloudfront examples

[mariandumitrascu-p]

#69

Correct and clean manifests and cloudfront examples

This pull request addresses some missing steps and correction in two examples of SDLF:

• Deequ and EMR steps using Step Functions (3)
• Manifest based processing (6)

Here is an aggregated summary of changes:

• all changes are contained in ./sdlf-utils/pipeline-examples/cloudfront and sdlf-utils/pipeline-examples/manifests folders.
• included steps for deploying the new pipelines, or datasets from local env, if a pipeline does not exists for them
• added requirements for running deployments from a local development machine
• replaced handling of a transformation through an extra branch (emr), since this would be harder to maintain in a multi deployment environment (dev, test prod). Replaced with creating its own stageB repo
• changes to maintain initial naming convention for transformations. emr is breaking the convention (example the convention is light_transform_<short-desc>.py. Cloudfront was using different one.
• added all missing steps in the correct order
• segregated submitting data, from deploying the glue job, or emr scripts in the last step. Users should be able to submit data multiple times without re-deploying anything.

This changes were extensibly tested by me and a team 3 developers from PREDICTif Solutions.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.