feat: add user-scoped memory extraction policies

[heaoxiang-ai]

Description

Add user-scoped memory extraction policies for enterprise multi-user OpenViking instances. User defaults are persisted in user_config.json, resolved at commit time, and frozen into the asynchronous extraction task so policy changes apply immediately without restarting the instance.

Human Involvement

• A human participated in the implementation or review loop
• This PR was generated entirely by AI agents without human participation in the loop

Related Issue

N/A

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Refactoring (no functional changes)
• Performance improvement
• Test update

Changes Made

• Extend MemoryPolicy with independent self/peer type filters, reject Agent memory for peers, and expand experiences to cases/trajectories/experiences only when resolving the effective Self policy.
• Resolve policy precedence as commit override > session metadata > latest user config > kernel default, apply the account Agent Evolution switch, and freeze the effective policy before the worker runs.
• Persist user defaults in viking://user/{user_id}/settings/user_config.json; add locked backup/update behavior and allowlisted GET/PATCH admin settings APIs.
• Propagate separate self/peer filters through the active v3 compressor and the memory isolation/streaming update path; leave the deprecated v2 compressor unchanged.
• Document the new admin APIs in English and Chinese and add policy, hot-update, and extraction-scope tests.

Testing

• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I have tested this on the following platforms:
  • Linux
  • macOS
  • Windows

Commands run:

• pytest -q -o addopts='' tests/unit/session/test_memory_policy.py tests/unit/session/test_agent_evolution_policy.py tests/server/test_admin_api.py::test_user_memory_policy_can_be_initialized_and_hot_updated — 16 passed.
• python -m compileall -q openviking — passed.
• ruff check for all changed Python and test files — passed.

The Session integration fixtures could not initialize in the local environment because native AGFS/VectorDB setup failed before reaching the test assertions (AGFSInvalidOperationError / local store lock). The added integration cases remain in the suite for CI.

Checklist

• My code follows the project's coding style
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

Screenshots (if applicable)

N/A

Additional Notes

• The dependent Serverless control-plane MR is https://code.byted.org/data-arch/knowledge_base_server/merge_requests/3405 and should be released after the kernel change.
• npm run check:api still reports the pre-existing missing reference for PATCH /api/v1/sessions/{}/config; the two routes introduced here are present in both overview and detailed API references.
• This PR intentionally does not add the optional batch Agent Evolution user list API; that endpoint requires a batch asset/policy contract to avoid N+1 settings requests.