GenAI memory ops extras (examples/prototypes)

[nagkumar91]

Companion to #3250. This PR contains the non-minimal changes split out (prototypes, examples, ancillary docs/workflows).

[nagkumar91]

Follow-up: cleaned the split so this draft PR now contains only the non-minimal extras (prototypes/examples/provider docs). All GenAI memory ops YAML/spec/generated docs/changelog remain in #3250.

[lmolkova]

please remove python code from semantic conventions, you can always publish it as a separate repo and provide link in the PR description

[github-actions]

This PR has been labeled as stale due to lack of activity. It will be automatically closed if there is no further activity over the next 7 days.

[nagkumar91]

Closing this PR — all relevant content has been consolidated into the non-normative implementation spec gist linked from #3250:

📄 Non-Normative Implementation Spec

The gist now includes:

• Full YAML spec (attributes, spans, metrics)
• 6 real-world stories with complete trace hierarchies (customer support, shopping assistant, multi-agent research, multi-tenant SaaS, compliance/debugging, GDPR lifecycle)
• LangChain framework mapping (ConversationBufferMemory, EntityMemory, VectorStoreRetrieverMemory → span mappings)
• Cross-framework mapping (Mem0, CrewAI, AutoGen, Letta/MemGPT)
• Rationale sections (memory vs retrieval, memory vs db conventions, why separate spans)

Implementation PRs:

• opentelemetry-python-contrib#4252 — Mem0
• traceloop/openllmetry#3713 — CrewAI