[prometheusexporter] Add job and instance labels to match federated prometheus

[dashpole]

Description
As described in the OpenTelemetry -> prometheus conversion specification, prometheus exporters must add job and instance labels to exported metrics in the collector:

In the collector Prometheus exporters, the service.name and service.namespace attributes MUST be combined as <service.namespace>/<service.name>, or <service.name> if namespace is empty, to form the job metric label. The service.instance.id attribute, if present, MUST be converted to the instance label; otherwise, instance should be added with an empty value.

This allows a user to scrape the prometheuxexporter's endpoint with honor_labels: true to preserve the original job and instance labels as described in https://prometheus.io/docs/prometheus/latest/federation/#configuring-federation.

Currently, the prometheus exporter only adds metric attributes:

opentelemetry-collector-contrib/exporter/prometheusexporter/collector.go

Lines 82 to 98 in 5cd8dd9

	func (c *collector) getMetricMetadata(metric pdata.Metric, attributes pdata.AttributeMap) (*prometheus.Desc, []string) {
	keys := make([]string, 0, attributes.Len())
	values := make([]string, 0, attributes.Len())
	attributes.Range(func(k string, v pdata.AttributeValue) bool {
	keys = append(keys, sanitize(k, c.skipSanitizeLabel))
	values = append(values, v.AsString())
	return true
	})
	return prometheus.NewDesc(
	c.metricName(c.namespace, metric),
	metric.Description(),
	keys,
	c.constLabels,
	), values
	}

[gouthamve]

Hi, I am working on this issue. Just to make sure there are not others working on it as well :)

[dashpole]

Hey @gouthamve, I see your comment on #5757 (comment). For manual testing of honor_labels, you use an upstream prometheus server. For unit testing, can we just make sure job and instance appear correctly on scraped metrics without using the prometheus receiver?