When hit max_output_tokens, some events are not emitted and logged

When the AI's output is truncated because it exceeds `max_output_tokens` and a subsequent request is made, both the current response and the next "continue" request are lost.

This happens in BYOK usage. I use the custom provider and it provides `65536` `max_output_tokens`. So it easily triggered this bug.

See this `events.jsonl` below (input and output is ommited for privacy):

```jsonl
{"type":"session.start","data":{"sessionId":"summary-qq-group-957302634-1788584545","version":1,"producer":"copilot-agent","copilotVersion":"0.0.0","startTime":"2026-09-05T05:02:26.005Z","selectedModel":"deepseek-v4-flash","reasoningEffort":"high","contextTier":null,"context":{"cwd":"/home/cola/Softwares/kanade-bot"},"remoteSteerable":false,"alreadyInUse":false},"id":"3e23bf11-4b01-4b8a-bc43-ad9b7d007c69","timestamp":"2026-09-05T05:02:26.024Z","parentId":null}
{"type":"user.message","data":{"content":"...","transformedContent":"...","messageId":"33fe414d-6631-449a-a351-3a34b2c2b68d","supportedNativeDocumentMimeTypes":[],"delivery":"idle","interactionId":"9596a41f-2bfc-4191-998c-740b2e30caec","turnId":"0","parentAgentTaskId":"65b27352-d09e-4c72-956b-b69594366d8c"},"id":"f56f57fc-8c67-4bd2-a372-cfb61f2c4109","timestamp":"2026-09-05T05:02:29.423Z","parentId":"3e23bf11-4b01-4b8a-bc43-ad9b7d007c69"}
{"type":"system.message","data":{"role":"system","content":"...","interactionId":"9596a41f-2bfc-4191-998c-740b2e30caec"},"id":"a33407ed-4a7b-4feb-ad75-83d6b0eac50b","timestamp":"2026-09-05T05:02:29.445Z","parentId":"f56f57fc-8c67-4bd2-a372-cfb61f2c4109"}
{"type":"assistant.turn_start","data":{"turnId":"0","interactionId":"9596a41f-2bfc-4191-998c-740b2e30caec"},"id":"5583fe72-b633-44ea-a183-7ac781d075d2","timestamp":"2026-09-05T05:02:29.458Z","parentId":"a33407ed-4a7b-4feb-ad75-83d6b0eac50b"}
{"type":"assistant.turn_end","data":{"turnId":"0"},"id":"5df48ed1-fe69-4ffe-8645-f0fa2545ef14","timestamp":"2026-09-05T05:03:40.349Z","parentId":"5583fe72-b633-44ea-a183-7ac781d075d2"}
{"type":"assistant.turn_start","data":{"turnId":"1","interactionId":"9596a41f-2bfc-4191-998c-740b2e30caec"},"id":"53c4e5ae-2365-4391-86b3-c39735958279","timestamp":"2026-09-05T05:03:40.349Z","parentId":"5df48ed1-fe69-4ffe-8645-f0fa2545ef14"}
{"type":"assistant.turn_end","data":{"turnId":"1"},"id":"d48f14f2-263b-4990-8d66-f5e858a49b10","timestamp":"2026-09-05T05:04:51.913Z","parentId":"53c4e5ae-2365-4391-86b3-c39735958279"}
{"type":"assistant.turn_start","data":{"turnId":"2","interactionId":"9596a41f-2bfc-4191-998c-740b2e30caec"},"id":"e77c5372-08d4-47ba-b886-5a75ae6a5a6d","timestamp":"2026-09-05T05:04:51.914Z","parentId":"d48f14f2-263b-4990-8d66-f5e858a49b10"}
{"type":"assistant.message","data":{"messageId":"51f06be7-30d8-4d28-b8ae-e8dc1fcc9937","model":"deepseek-v4-flash","content":"（继续）...","toolRequests":[],"interactionId":"9596a41f-2bfc-4191-998c-740b2e30caec","turnId":"2","reasoningText":"Let me continue writing the topic summary. I was in the middle of topic 12. Let me continue from there, then complete the personal image section.","rte":false,"apiCallId":"6197b23f-7420-44dd-8bc4-a032e5784de2"},"id":"7fd35caa-4f74-4365-84d5-c39f7556561a","timestamp":"2026-09-05T05:05:41.237Z","parentId":"e77c5372-08d4-47ba-b886-5a75ae6a5a6d"}
{"type":"assistant.turn_end","data":{"turnId":"2"},"id":"1d7b9c22-2a37-4a73-b54e-f7f398c00f40","timestamp":"2026-09-05T05:05:41.244Z","parentId":"7fd35caa-4f74-4365-84d5-c39f7556561a"}
{"type":"session.shutdown","data":{"shutdownType":"routine","totalPremiumRequests":0,"totalNanoAiu":0,"totalApiDurationMs":190704,"sessionStartTime":1788584546005,"eventsFileSizeBytes":368878,"codeChanges":{"linesAdded":0,"linesRemoved":0,"filesModified":[]},"modelMetrics":{"deepseek-v4-flash":{"requests":{"count":3,"cost":0},"usage":{"inputTokens":234114,"outputTokens":21467,"cacheReadTokens":147456,"cacheWriteTokens":0,"reasoningTokens":14945},"totalNanoAiu":0}},"agentMetrics":{"main":{"totalApiDurationMs":190704,"totalNanoAiu":0.0,"modelMetrics":{"deepseek-v4-flash":{"requests":{"count":3,"cost":0.0},"usage":{"inputTokens":234114,"outputTokens":21467,"cacheReadTokens":147456,"cacheWriteTokens":0,"reasoningTokens":14945},"totalNanoAiu":0.0}}}},"currentModel":"deepseek-v4-flash","currentTokens":81090,"systemTokens":816,"conversationTokens":68907,"toolDefinitionsTokens":11364},"id":"391655ed-981c-4225-b84c-8948504c7bae","timestamp":"2026-09-05T05:05:41.537Z","parentId":"1d7b9c22-2a37-4a73-b54e-f7f398c00f40"}
```

OS: Linux Mint 22.3
Copilot SDK Version: v1.0.13
Copilot CLI Version: v1.0.83 (pinned version)

No files, tests, or entry points are identified. Start by reproducing a BYOK request that reaches max_output_tokens, then trace event emission and logging across the truncated response and subsequent continue request. Done means both responses are present in events.jsonl and are logged consistently.

索引モデルが issue の本文から書いたものです。