This PR contains preliminary functionality for writing feed items as complete documents to S3, closing #5.

Done

[x] Created a new organization in S3 with various IAMs for app development
[x] Created a bucket for baleen ingestion and set up all the credentials as local environment vars
[x] Added a store module that has structs for AWS credentials and for documents to be written to S3
[x] Handled irregular feeds including those that don't have the correct date (since we're using date to determine filenames in s3) and those that have malformed XML
[x] Implemented preliminary functionality inside fetch module to retrieve full text of articles from XML URL provided by feed (resolves #7)
[x] Added a levelDB manifest (resolves #8) so we don't re-fetch documents we already have

Proposed follow-on tasks

validate encoding (see #9)
upgrade configuration methodology (more like what we're doing for the messaging backend?)
put script on a schedule (currently requires a person to kick it off) & deploy as a lambda?
other things?