Parsedmarc

-----

- Use `publixsuffix2` (closes issue 4)
- Add Elasticsearch to automated testing
- Lock `elasticsearch-dsl` required version to `6.2.1` (closes issue 25)

-----

**Note**: Re-importing `kibana_saved_objects.json` in Kibana [is required](https://domainaware.github.io/parsedmarc/#upgrading-kibana-index-patterns) when upgrading to this version!

- Bugfix: Reindex the aggregate report index field `published_policy.fo`
as `text` instead of `long` (Closes issue 31)
- Bugfix: IDLE email processing in Gmail/G-Suite accounts (closes issue 33)
- Bugfix: Fix inaccurate DNS timeout in CLI documentation (closes issue 34)
- Bugfix: Forensic report processing via CLI
- Bugfix: Duplicate aggregate report Elasticsearch query broken
- Bugfix: Crash when `Arrival-Date` header is missing in a
forensic/failure/ruf report
- IMAP reliability improvements
- Save data in separate indexes each day to make managing data retention easier
- Cache DNS queries in memory

-----

- Don't crash if Elasticsearch returns an unexpected result (workaround for issue 31)

-----

- Packaging fixes

-----

- Kafka output improvements
- Moved some key values (`report_id`, `org_email`, `org_name`) higher in the JSON structure
- Recreated the `date_range` values from the ES client for easier parsing.
- Started sending individual record slices. Kafka default message size is 1 MB, some aggregate reports were exceeding this. Now it appends meta-data and sends record by record.

-----

- Fix decoding of attachments inside forensic samples
- Add CLI option `--imap-skip-certificate-verification`
- Add optional `ssl_context` argument for `get_dmarc_reports_from_inbox()`
and `watch_inbox()`
- Debug logging improvements