feat: Refactor loongsuite-instrumentation-dify for the latest OTel/LoongSuite semantic conventions.

[Cirilla-zmh]

Semantic conventions:

• OTel GenAI semconv: https://github.com/open-telemetry/semantic-conventions/blob/main/docs/gen-ai/gen-ai-spans.md
• OTel GenAI Agent semconv: https://github.com/open-telemetry/semantic-conventions/blob/main/docs/gen-ai/gen-ai-agent-spans.md
• Entry Span semconv (experimental): Enhancing OTel GenAI Semantic Conventions with a User-Centric Entry Span open-telemetry/semantic-conventions-genai#302
• ReAct Span semconv (experimental): Adding ReAct Iterations Spans in Reasoning-Acting Agents open-telemetry/semantic-conventions-genai#81

[ralf0131]

Issue Evaluation

Category: type/enhancement (refactor) | Status: Evaluated

Alignment work to keep loongsuite-instrumentation-dify consistent with the latest OTel/LoongSuite semantic conventions.

Feasibility: feasible — depends on the delta between the current dify instrumentation and the target semconv/util shape.
Scope: loongsuite-instrumentation-dify only; should reuse opentelemetry-util-genai helpers for attribute mapping and token metrics.
Compatibility: primarily a refactor; verify no public API breakage in the difi instrumentation package.

For a concrete plan, it would help to enumerate which specific attributes/events currently drift from the target semantic conventions.

---

Automated evaluation by github-manager-bot