fix(gateway): separate text and media lanes in AI analysis queue

Image messages previously blocked the whole analysis pipeline:
- conversationProcessing was a single lock per conversation; processBatch
  awaited BOTH text and media worker jobs before releasing it, so a fast
  text verdict sat unused until the slow vision/media batch finished
- one global LLM semaphore (AI_LLM_MAX_CONCURRENT) was shared by text and
  media, so a vision backlog could starve text inference
- recovery worker gated on conversationProcessing.size

Now the queue is split into independent text/media lanes:
- conversationProcessing maps key -> Partial<Record<lane, startedAt>>;
  each lane holds its own lock and frees it the moment ITS worker job
  resolves (ownership-guarded clear prevents stale timers clearing newer
  slots)
- two LLM semaphores: AI_LLM_MAX_CONCURRENT (text, default 8) and
  AI_LLM_MEDIA_MAX_CONCURRENT (media, default 4) via
  withLlmConcurrency(fn, { lane })
- batchScheduler schedules per conversation+lane (timer keys
  '<key>::<lane>'); splitMessagesByLane/laneOfMessage moved to pure
  analysisLanes.ts (unit-testable without Piscina)
- ai-analysis-worker batch jobs carry a lane field; per-lane active
  request gauges (active_text_requests / active_media_requests)
- added tests/analysisLaneLock.test.ts (7 tests: independent lane locks,
  preserving other-lane lock, clear-all, ownership guard, lane split)

Docs: ARCHITECTURE.md + AGENTS.md concurrency model updated.
typecheck/lint/test(138)/build all green.

services/discord-gateway/AGENTS.md

@@ -69,8 +69,14 @@ textBatchProcessor.ts    mediaBatchProcessor.ts         llmClient.ts
 ```
 
 - Entry: `aiAnalyzer.ts` (`queueMessageAnalysis`, `startPendingAIAnalysisWorker`)
-- Concurrency: LLM semaphore (`AI_LLM_MAX_CONCURRENT`, default 5)
+- Concurrency: **two per-lane LLM semaphores** (2026-09-24) — text
+  (`AI_LLM_MAX_CONCURRENT`, default 8) and media/vision
+  (`AI_LLM_MEDIA_MAX_CONCURRENT`, default 4); a media backlog can never
+  consume text slots
 - Piscina: text pool (4 threads) + media pool (2 threads)
+- Locks are **per conversation per lane** (`conversationProcessing` maps key →
+  lane → startedAt): the text lane of a conversation never waits on that
+  conversation's media lane (this was the "image blocks the queue" bug)
 - **Each worker thread has its own pg Pool** (min 0, grows to `POSTGRES_POOL_MAX`)
 
 ## Module: message-capture