docs: link distributed tracing extension to otel semconv

[joaopgrassi]

Replacement for #912

Proposed Changes

This follows the same idea/goal as #912 - Point to the OpenTelemetry CloudEvents semantic conventions to help people to understand how to use the extension and how a trace structure should look like in systems using CloudEvents.

I took a slight different approach with this PR where I left most of the existing text. We have now finalized the CloudEvents semconv in OTel. The group there decided to only mention the distributed tracing extension as an additional mechanism to propagate context, but the definition of it should stay here. OTel also doesn't document/specify other means of propagation in messaging systems for example, we rather only mention it. That's why we followed the same for CloudEvents.

The document now points to the CloudEvents OTel semconv which basically tells to follow the messaging semantic conventions.

Release Note

Clarify usage of Distributed Tracing Extension by OpenTelemetry

Note: The PR #912 is very old and I had many conflicts trying to rebase it. So I'm opening this new one. CC @duglin

Note 2: The link to the OTel website doesn't contain the last version of the CloudEvents semantic conventions. The version that will go live soon is here: https://github.com/open-telemetry/semantic-conventions/blob/main/docs/cloudevents/cloudevents-spans.md

[jskeet]

I don't know much about the accuracy of the statements here, but they certainly seem useful.

[manuelottlik]

I just read through the website, the changes and the discussion in the preceding PR – this looks like a very useful extension. Just one question (that should not block this PR): Is this actually implemented in the most common SDKs already or is it too early for such widespread adoption?

[joaopgrassi]

Is this actually implemented in the most common SDKs already or is it too early for such widespread adoption?

@manuelottlik I implemented an "old" version of the OTel messaging conventions in the Go SDK some years ago, but I assume it's definitely not widespread implemented. Due to the conventions being in the works, there was no much point in working on it. Now that it is approaching stability it is a good time to start :)

[duglin]

Approved on the 7/25 call
thanks @joaopgrassi for all your work on this!