Most or all metrics exported by Retina should be counters, not gauges

[andreev-io]

Describe the bug
Retina is currently exporting gauge-type metrics in its modules (e.g. forward.go, drops.go. Since these metrics only go up (or are reset to 0), they should be counters. This is important since a metric type different from the metric's semantical meaning can break the implicit contract and assumptions that people make when querying the data.

From Prometheus documentation:

A gauge is a metric that represents a single numerical value that can arbitrarily go up and down.

A counter is a cumulative metric that represents a single monotonically increasing counter whose value can only increase or be reset to zero on restart. For example, you can use a counter to represent the number of requests served, tasks completed, or errors.

To Reproduce
N/A

Expected behavior
N/A

Screenshots
N/A

Platform (please complete the following information):
N/A

Additional context
N/A

[andreev-io]

I'm going to open a PR to address this in the forward plugin as an example.

[andreev-io]

@rbtr Could you assign this to me? I opened #238, but I need some help with figuring out how to build Retina locally: #238 (comment).

[nddq]

just look at these metrics recently and thought of this issue. The reason why we are doing gauges and not counters, even though these metrics supposed to be only ever increasing, is that because we are doing snapshots of the bpf map periodically and Set the metric to whatever the values we've read. The actual counting is being done in eBPF. We should rename <MetricName>Counter to <MetricName>Gauge to prevent confusion