feat(openviking): add full recall prefetch policy (salvage #48927)

[kshitijk4poor]

Summary

Salvage of #48927 (consolidating OpenViking recall work from #41706, #33260, #49975, #32444). Replaces stale background post-turn prefetch warming with synchronous current-query recall — the old queue_prefetch warmed the PREVIOUS user message while prefetch() consumed the CURRENT one, so injected context was always about the wrong topic.

Changes

• plugins/memory/openviking/__init__.py: prefetch() now does session-aware /api/v1/search/search with the current query, falls back to /api/v1/search/find on failure. Contract-safe payloads (limit, score_threshold, context_type, session_id) — no top_k, no search-body mode, no target_uri. L2 content reads for items with level=2 or empty abstracts, capped at full_read_limit. Local ranking, dedup, score threshold, injected-char budget. queue_prefetch() is now a no-op. Additive batched viking_read (uris param, up to 3). Per-request timeout on _VikingClient.get/post/delete. Removes stale _prefetch_* state. Strengthened system_prompt_block.
• tests/openviking_plugin/test_openviking.py: 43 tests covering recall config, session-aware search, contract-safe payloads, L2 reads, budget caps, batched reads, E2E local HTTP server.
• tests/plugins/memory/test_openviking_provider.py: 133 tests covering session search, fallback, no top_k/mode/target_uri, L2 reads, empty-abstract fallback, no-op queue_prefetch, bounded timeouts, full-read caps.

Salvage follow-up fixes (vs original PR #48927)

1. Exposed all 8 recall config knobs in get_config_schema() — PR feat(openviking): add full recall prefetch policy #48927 had removed them; Openviking recall policy #41706 correctly exposed them. Env vars remain as internal mechanism but are now visible in setup wizard.
2. Lowered default timeout 8s→4s, request_timeout 6s→3s, full_read_limit 3→2 — reduces per-turn blocking latency (synchronous prefetch runs on the hot path before every LLM call).

Validation

Test suite	Before	After
tests/openviking_plugin/test_openviking.py	—	43 passed
tests/plugins/memory/test_openviking_provider.py	—	133 passed
tests/agent/test_memory_*.py (regression)	—	122 passed
E2E (isolated HERMES_HOME, real imports)	—	6/6 passed

Attribution

• @ehz0ah (Hao Zhe) — PR feat(openviking): add full recall prefetch policy #48927, commits 102c8d3, c6fa722, 6f0b5e0, 60f8607
• @huangxun375-stack (Eurekaxun) — PR Openviking recall policy #41706, commit c982d94 (cherry-picked into feat(openviking): add full recall prefetch policy #48927)
• @per1970 — PR fix(memory): synchronous OpenViking prefetch with current turn's query #33260 (original synchronous-prefetch diagnosis)
• @wgd753 — PR fix(openviking): sync prefetch with current query + use limit #49975 (limit vs top_k, prefetch timeout)

Closes #48927, #41706, #33260, #49975, #32444