Prometheus metrics for channels

[alexandrosfilios]

We need a consistent way of monitoring how full the buffered channels of the application are. We need to keep track of len(ch)/len(cap) to provide this data to Grafana for visualization. This way, it will be easier to identify bottlenecks.

[adecaro]

All right, if I understand correctly, we need to do the following:

• List of the channels that we use.
• Introduce a reference to metrics.Provider in each service that contain channels
• At channel creation time, we signal the capacity of the channel to the metrics.
• When an element is read/removed from the channel, we signal this to the metrics.

@alexandrosfilios , does this match what you had in mind.