feat: add support to elasticsearch v2.x

[dougppaz]

• add elasticsearch 2.4.6 nodestats fixture
• load node stats os.load_average as os_load1 gauge
• load node stats os.cpu_percent as os_cpu_percent gauge
• add indice suggest (total, time and current) from node stats

[SuperQ]

In general, we don't expose OS stats from systems like this. Since they overlap with metrics from the node_exporter or cAdvisor, depending on the runtime. While things like ElasticSearch expose these metrics, we don't forward them on to Prometheus.