Tempo Metrics Generation uses different metrics labels to Grafana Agent

[hedss]

Is your feature request related to a problem? Please describe.
In Grafana Agent, the generated metrics labels are:

traces_spanmetrics_latency_sum
traces_spanmetrics_latency_count
traces_spanmetrics_latency_bucket
traces_spanmetrics_call_counts

In Tempo metrics server side we generate metrics called:

traces_spanmetrics_duration_seconds_sum
traces_spanmetrics_duration_seconds_count
traces_spanmetrics_duration_seconds_bucket
traces_spanmetrics_call_counts

Obviously, the first three differ between the Agent and Tempo.

There is also a difference on the label for the status code, in Grafana Agent it's:

status_code

and in Tempo metrics generation it's:

span_status

Describe the solution you'd like
We should either consider altering these to the OTel standard, or at least documenting these differences so that should users change from Grafana Agent metrics generation to Tempo, they understand the changes they will need to make.

Describe alternatives you've considered
N/A

Additional context
N/A

[knylander-grafana]

@hedss On the documentation side, is this something that might be handled by a UI error message warning the user?

[hedss]

@knylander-grafana That's an interesting question, but Grafana would probably need to understand that trace span metrics are being received and the user's not requesting the right ones, so it'd need to look at panels using those names (or the Explorer). Not sure how viable this is.