Prometheus-es-exporter

Fixes:
- Replace syntax that is incompatible with Python 3.4

Fixes:
- Allow changing the timeouts used when fetching cluster metrics
- Log cluster metric fetch timeouts as warnings, rather than errors
- Report whether a cluster metric fetch succeeded via a metric, [see here for why](https://prometheus.io/docs/instrumenting/writing_exporters/#failed-scrapes)

Breaking Changes:
- Default port changed from 8080 to 9206, [see here for why](https://github.com/prometheus/prometheus/wiki/Default-port-allocations)
- Cluster metrics queried from Elasticsearch on metric scrape, not on a schedule, [see here for why](https://prometheus.io/docs/instrumenting/writing_exporters/#scheduling) (Data queries are still run on schedules)

Enhancements:
- Allow retrieving fieldata info for specific fields
- Expose query `took` time as a metric
- Allow limiting metrics returned in nodes/indices stats
- Allow enabling SSL verification (thanks echojc)
- Allow queries to have custom timeouts (thanks echojc)
- Allow setting arbitrary logging levels

Enhancements:
- Add cluster metrics

Breaking Changes:
- New bucket label format

Enhancements:
- Add SIGTERM handler
- Optimise Dockerfile
- Add JSON structured logging support

Bug Fixes:
- Fix exceptions when generating label values for numeric bucket names