feat: native short-term memory + everos long-term consolidation

[YueAWu]

Change description

Restructures the EverOS-backed memory subsystem so native memory owns short-term and EverOS owns long-term, and ports three recall improvements from the legacy everclaw repo.

Ported from the legacy repo (env vars renamed EVERCLAW_* -> RAVEN_*):

• Drop current-session hits from the # Memory recall so it does not duplicate the live conversation already in the history window (over-fetch top_k*2, then filter by session_id).
• raven agent --message-file for prompts beyond the shell ARG_MAX.
• Env-gated episodic recall + preference annotation in the consolidator (RAVEN_EPISODIC_TOPK / RAVEN_ANNOTATE_PREFS / RAVEN_EPISODIC_CLUSTER), default-off.

EverOS call-timing refactor (native = short-term, EverOS = long-term):

• Recall is now on-miss: native runs every turn; EverOS user-track recall fires only when native surfaces nothing relevant to the query.
• The per-turn EverOS store is removed. EverOS consolidation runs only (a) intra-day, in the background, when a session crosses everos_consolidation_threshold_pct of the context window (default 80), and (b) nightly at everos_nightly_consolidation_hour (default 0, offline).
• backend.store gains force_flush to promote a batch immediately; consolidation is incremental per session (a metadata watermark) so the intra-day and nightly triggers never double-ingest.
• Native short-term compaction (maybe_consolidate_by_tokens) is unchanged.

Testing

• Unit: threshold / nightly / on-miss paths and a full in-memory lifecycle smoke (test_agent_loop_everos_pipeline.py), plus updated test_segments.py / test_default_context_engine.py for the new over-fetch behavior.
• Real EverOS e2e (test_everos_backend_e2e.py, real_llm): existing dual-track recall plus a new force_flush single-store recall.
• Full unit sweep shows zero net regression versus a stashed baseline.
• Not run: a full gateway-boot CLI smoke (no provider/runtime e2e here).

Type of change

• Bug fix
• New feature
• Document
• Others

Related issues (if there is)

None.

Checklists

Development

• Lint rules pass locally
• Application changes have been tested thoroughly
• Automated tests covering modified code pass

Security

• Security impact of change has been considered
• Code follows security best practices and guidelines

Code review

• Pull request has a descriptive title and context useful to a reviewer. Screenshots or screencasts are attached as necessary