New changeset parsing pipeline

[sunu]

We need to change the pipeline to adapt and get the data from other sources because the Mapbox bucket that used to store augmented diffs is not being updated anymore.

With this new script:

• everything other than the final geojson to FGB conversion is in a single nodejs script; trying to reduce the clutter that comes from combines multiple different node scripts and bash scripts
• The pipeline grabs hourly replication files from https://planet.openstreetmap.org/replication/hour/
• It parse the xml in the hourly replication file as a stream and creates a set of all unique changeset ids mentioned in the replication file
• For each of the collected changeset ids, it fetched the "real changeset" json which contains the added/removed/modified features. These changesets are fetched from the real-changesets s3 bucket (https://real-changesets.s3.amazonaws.com/<changeset-id>.json)
• Then we combine the features from each of these changesets into a single valid geojson file
• In a later step in the pipeline we will convert this combined geojson file into an FGB and upload to cloud storage.

@kamicut @willemarcel Please take a look and let me know if anything looks odd