feat: opentelemetry-collector-observ-lib

[theSuess]

This PR adds a new observ-lib for the OpenTelemetry internal metrics.

Screenshot of the overview dashboard:

[cla-assistant]

All committers have signed the CLA.

[theSuess]

TBH not sure where the diff in this file is coming from. My changes shouldn't casue this but needed to regenerate the dashboard for CI to pass

[cyrille-leclerc]

It's great!
Nothing blocking in this feedback.

Additional question: do we plan to add Loki logs?

[theSuess]

I'll elaborate on some decisions mentioned in the feedback:

• The reason behind breaking down processors by component is that it doesn't make sense to aggregate them as chains of processors lead to sums that aren't useful. For receiving and exporting, the aggregation is a useful statistic
• Good catch on the Service_instance_id and job field. I was working based on the singals lib which is a bit biased towards prometheus. Will fix this to use service_name instead of job (same value, gets mapped by the otel endpoint)
• People will use the dashboard outside of k8s as well so it's not included specifically anywhere. I'll look into adding the host name as we'll have that information in most situations

Which other KPIs would you consider important for the collectors?