RemoteA2aAgent: streamed A2A chunks and the final replacement event get different Event ids, so answer can be shown twice

Finding: In google-adk 2.11.0 and main at commit 63aed55 (reported on macOS, Python 3.12), RemoteA2aAgent consuming an A2A streamed answer gives every chunk event and the final replacement event a different Event.id, so a consumer that appends text per event can display the answer twice ([issue](https://github.com/google/adk-python/issues/7395)). This is an external report and was not reproduced by KFA. Reported details: the remote side uses A2aAgentExecutor with SSE. The reporter says the A2A server marks the final event correctly: the first chunk has append=False, later chunks append=True, and the last has last_chunk=True with append=False, meaning replace ([issue](https://github.com/google/adk-python/issues/7395)). On the client, the v2 converter and the legacy handler reportedly translate only last_chunk into partial, create each Event without an id so a new one is minted, and drop append and the artifact id. Nothing then links the final event to its chunks. The reporter contrasts this with local BaseLlmFlow streaming, where partial chunks and the closing complete event share an id. Their sample output shows four text events with four distinct ids. They report that ADK 2.7.1 behaves the same, so it is not described as a regression ([issue](https://github.com/google/adk-python/issues/7395)). Status: a responder commented on 2026-10-07 that this is confirmed on main, that both client paths omit the id and drop artifact_id, and that deriving the id from invocation_id and artifact_id is the right direction. Review continues in PR #7396 ([comment](https://github.com/google/adk-python/issues/7395#issuecomment-6031473705)). The commenter's association is NONE, and no merge, fix commit or release is stated in the supplied sources, so fix status is unknown/proposed. Consumer-side mitigation reported by the contributor: keep the text already sent for the current answer, and for each non-partial event emit only the part beyond that text, resetting when a tool result ends the step. The contributor says this can wrongly merge distinct messages when new text starts like the old text, for example Done. followed by Done. Next step ([comment](https://github.com/google/adk-python/issues/7395#issuecomment-5971180518)). Treat it as a heuristic, not a fix. Limitations: the repro script was not executed here, PR #7396 was not inspected, and the available comment history may be incomplete.