feat(harness): background drain ingestion, auto-context snapshot cache, and configurable reasoning-effort policy

[LiuStar656]

feat(harness): background drain ingestion, auto-context snapshot cache, and configurable reasoning-effort policy

Problem

The DeepSeek Harness integration ingests memory synchronously: every completed
agent turn immediately runs StrataGate's LLM memory work (appendTurn, block
sealing, activated-memory retrieval) inline on the hot path, so conversation
assembly blocks on memory processing. The structured-call reasoning-effort
policy is also not user-configurable: force-off raises a desktop
notification and fails the structured call on providers that do not advertise
off.

Solution

Three commits, one concern each:

1. da9ea51 — Background drain + snapshot cache (src/runtime.ts, src/index.ts, tests).
   Turns are folded on the hot path then only counted and queued; the actual
   LLM memory work runs on an unref()'d timer with per-session
   serialization, count-triggered eager draining (≥3 queued turns → 150 ms),
   and exponential backoff on failure. The system-prompt/assemble hook reads
   a synchronous snapshot cache refreshed by the drain instead of doing
   flush + retrieval inline; an empty or stale snapshot is skipped, or carries
   an explicit staleness marker. flush() stays the explicit synchronous
   point; a public refreshAutoContext(session) provides a deterministic
   refresh for tests and admin flows.
2. 08600b8 — Effort degrade instead of failure (src/llm.ts, src/config.ts, tests).
   When the provider does not advertise off, force-off now falls back to
   the model default with a single warning instead of failing the structured
   call. The resolved config field is optional (default auto), so existing
   plugin configs and tests stay valid.
3. a31d7c6 — Effort policy setting (src/client.js, src/index.ts, package.json, CHANGELOG).
   New @deepseek-ai/dsh-settings wiring: the effort policy (auto | force-off) is a user-editable switch on the plugin's settings page, bound
   through the client settings.section registration
   (register.inject carries hooks/setEffort/resetEffort) and mirrored into
   the model bridge on every change, so the current mode is read per structured
   call. Adds the peer/dev dependency and lockfile update; CHANGELOG gains the
   Unreleased entries.

Acceptance

• npm run check --workspace stratagate-dsh — zero type errors.
• npm run test --workspace stratagate-dsh — 59/59 passing (8 files),
  including new background-drain/snapshot-cache tests and the adapted
  plugin-composition test (assemble skips the memory context until the drain
  has rendered a snapshot).
• Hot-path assemble is a synchronous cache read: no IO, never throws.
• Backwards compatible: flush() still forces queued turns; config defaults
  unchanged; upstream multibatch protocol and turn-stopping are preserved.

---

中文说明

问题

DeepSeek Harness 集成的记忆摄取是同步的：每次回合结束都立即在热路径上运行
StrataGate 的 LLM 记忆处理（appendTurn、封存、激活记忆检索），对话组装会被
记忆工作阻塞；推理档位策略也不可配置，force-off 在模型不支持 off 时会发
桌面通知并使结构化调用失败。

方案（三个 commit，各管一事）

1. da9ea51 后台计数驱动摄取 + 快照缓存（src/runtime.ts、src/index.ts、测试）。
   回合在热路径只折叠入队并计数；真正的 LLM 记忆工作在 unref() 定时器上执行，
   按会话串行，批量压力触发 eager drain（队列 ≥3 回合 → 150ms），失败指数退避。
   system-prompt/assemble 钩子改为读取后台 drain 渲染的同步快照缓存，空/过期
   快照直接跳过或带显式过时标记，热路径永不阻塞。flush() 仍是显式同步点；
   新增公开 refreshAutoContext(session) 供测试与管理流程确定性刷新。
2. 08600b8 降级代替失败（src/llm.ts、src/config.ts、测试）。
   提供方不支持 off 时，force-off 退回模型默认档位并告警一次，不再失败。
   resolved 配置字段改为可选（默认 auto），现有插件配置与测试无需改动。
3. a31d7c6 档位设置化（src/client.js、src/index.ts、package.json、CHANGELOG）。
   新接入 @deepseek-ai/dsh-settings：档位策略（auto | force-off）成为插件
   设置页的用户可编辑开关，经客户端 settings.section 注册
   （register.inject 携带 hooks/setEffort/resetEffort）注入记忆 UI，改动即时
   镜像到模型桥，每次结构化调用按当前档位执行。新增 peer/dev 依赖与 lockfile
   更新；CHANGELOG 补充 Unreleased 条目。

验收

• npm run check --workspace stratagate-dsh：零类型错误。
• npm run test --workspace stratagate-dsh：59/59 通过（8 个文件），
  含新增的后台 drain/快照缓存测试与适配后的插件组装测试
  （assemble 在 drain 渲染出快照前跳过记忆上下文）。
• 热路径 assemble 为同步缓存读：无 IO、永不 throw。
• 向后兼容：flush() 仍强制消费积压回合；配置默认值不变；
  保留上游 multibatch 协议与 turn-stopping 行为。