Split multi use gen-ai spans in to 2 distinct spans #2467

[thompson-tomo]

Progresses #2467

Changes

In a couple of scenarios spans either had 2 name definitions or 2 span kinds. Each of these were for different use cases, hence these have been split into 2 spans.

Reason for the change

This is a blocker for moving to code generated span registries as a span can only have 1 type, hence should be separate definitions.

Note: if the PR is touching an area that is not listed in the existing areas, or the area does not have sufficient domain experts coverage, the PR might be tagged as experts needed and move slowly until experts are identified.

Merge requirement checklist

• CONTRIBUTING.md guidelines followed.
• Change log entry added, according to the guidelines in When to add a changelog entry.
  • If your PR does not need a change log, start the PR title with [chore]
• schema-next.yaml updated with changes to existing conventions.

[thompson-tomo]

This pr has been split to only include the gen-ai part. @open-telemetry/semconv-genai-approvers can you review

[lmolkova]

I don't believe that splitting spans by kind improves end user or instrumentation author experience. Now we duplicate huge span tables and the only difference between them is the kind.

Moreover, we know for a fact that client/internal is frequently a preference - e.g. Spring AI uses internal kind for external model calls - spring-projects/spring-ai#1174, more context #1315

[thompson-tomo]

So i have fixed the issue introduced by rebasing as identified by @gyliu513 and the internal span now no longer has the server attributes with the logic being if you know that you are talking to a sever then you aren't internal. Is there any other attributes which only apply when using a server based gen-ai model? For instance are usage tokens still applicable?

@lmolkova i am aware of the ambiguity around span king usage etc and believe that is addressed by using the phrase Should rather than must. I think this scenario is different as the documentation is clearly talking about 2 different kinds of activities, hence splitting them and now removing the non applicable server attributes. If there is others happy to move them to create a bigger difference in definitions.

[lmolkova]

I still don't understand which problem this PR is trying to solve and how is it solved by introducing embedded inference span in addition to inference and embeddings. I'm going to close it.