feat(memory): integrate multimodal video memory

[openjiuwen-sync-bot]

Paired: GitHub #126 ↔ GitCode !201

Summary

• integrate asynchronous video ingest through Control
• build CLM/ELM MemoryUnits through the native construction path
• retrieve native, CLM and ELM memories with RRF fusion

Validation

• 54 targeted and regression tests passed
• single commit with no documentation changes

Linked Closing Issues:

• Fixes [Feature]: 多模态记忆的处理技术和精度提升 #78

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

chenzhwei seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[openjiuwen-collaboration-bot]

head_sha: 30e7b2bfc721e111711e1b118b5ecc347e032078

🤖 正在生成合并请求摘要，请稍候…

[openjiuwen-collaboration-bot]

head_sha: 30e7b2bfc721e111711e1b118b5ecc347e032078

🤖 AI 代码检视正在进行中，请稍候…

[openjiuwen-collaboration-bot]

head_sha: 30e7b2bfc721e111711e1b118b5ecc347e032078

⚠️ 本次变更过大（23 个可审文件、6157 行，预估需 6 轮分片审查），已超出 AI 代码评审的处理范围，本次跳过。建议拆分为更小的 PR 以获得有效评审。

[openjiuwen-collaboration-bot]

head_sha: 30e7b2bfc721e111711e1b118b5ecc347e032078

⚠️ 本次变更过大（23 个可审文件、6157 行，预估需 6 轮分片审查），已超出 AI 代码评审的处理范围，本次跳过。建议拆分为更小的 PR 以获得有效评审。

[openjiuwen-collaboration-bot]

head_sha: 30e7b2bfc721e111711e1b118b5ecc347e032078

任务名称	结果	日志操作
静态检查	❌FAILED	点此跳转
禁用词扫描	N/A	N/A
防投毒检查	✅SUCCESS	点此跳转
开源合规检查	✅SUCCESS	点此跳转
UT测试	✅SUCCESS	点此跳转
ST测试	N/A	N/A
build 编译包	N/A	N/A
ruff codecheck	✅SUCCESS	N/A

[openjiuwen-collaboration-bot]

head_sha: fea3bdef94528821606e6b0b03f66e55c74d393f

任务名称	结果	日志操作
静态检查	✅SUCCESS	点此跳转
禁用词扫描	N/A	N/A
防投毒检查	✅SUCCESS	点此跳转
开源合规检查	✅SUCCESS	点此跳转
UT测试	✅SUCCESS	点此跳转
ST测试	N/A	N/A
build 编译包	N/A	N/A
ruff codecheck	✅SUCCESS	N/A