Missing pyrra_availability for latencyNative SLO

[samwright]

The latencyNative SLO indicator currently doesn't support generic/grafana rules. I've created a PR adding support for the pyrra_availability metric. It's a small change, guided by the grafana docs on using native histograms.

I haven't added the pyrra_errors_total or pyrra_requests_total rules because I don't think it is possible to generate a counter metric like those from a native histogram. The histogram contains values that act as counter metrics, which must go through a function like rate to be converted to values that act as a gauge metrics, and only then can the histogram_* functions be called on them. I think the best we can do is create a new gauge metric representing the requests + errors in the last e.g. 30s. This is annoying when creating grafana graphs, as zooming out means you might miss a peak...