Enable pushing metrics to influx as an alternative to pulling to prometheus

[evan-forbes]

          Instead of duplicating ourselves, can we implement the `metrics.Gauge` and `metrics.Counter` interfaces using this eventCollector and then in the `metrics.go` file of each reactor, include a `InfluxDBMetrics` function alongside the `PrometheusMetrics` function. That way an operator can choose to trace metrics using either prometheus or influxDB (perhaps in the future we can add support for both).

Originally posted by @cmwaters in #970 (comment)