docs: enterprise refactor plan with ralph specs

.ralph/specs/phase-08-job-queue.md

@@ -0,0 +1,66 @@
+# Phase 8: Job Queue System
+
+## Tabla jobs (SQLite)
+```sql
+CREATE TABLE IF NOT EXISTS jobs (
+  id TEXT PRIMARY KEY,
+  type TEXT NOT NULL,
+  status TEXT NOT NULL DEFAULT 'pending',
+  payload TEXT NOT NULL,
+  result TEXT,
+  error TEXT,
+  attempts INTEGER NOT NULL DEFAULT 0,
+  max_attempts INTEGER NOT NULL DEFAULT 3,
+  priority INTEGER NOT NULL DEFAULT 0,
+  run_at TEXT NOT NULL,
+  started_at TEXT,
+  completed_at TEXT,
+  created_at TEXT NOT NULL,
+  updated_at TEXT NOT NULL
+);
+CREATE INDEX IF NOT EXISTS idx_jobs_poll ON jobs(status, run_at, priority DESC);
+```
+
+## Interface
+```typescript
+export interface IJobQueue {
+  enqueue<T>(type: string, payload: T, opts?: { runAt?: Date; priority?: number; maxAttempts?: number }): Promise<string>;
+  start(): void;
+  pause(): void;
+  waitForActive(timeoutMs: number): Promise<void>;
+}
+```
+
+## Polling logic
+```
+loop (cada pollIntervalMs):
+  SELECT id, type, payload FROM jobs
+  WHERE status = 'pending' AND run_at <= datetime('now')
+  ORDER BY priority DESC, created_at ASC
+  LIMIT 1
+
+  if found:
+    UPDATE jobs SET status = 'running', started_at = now, attempts = attempts + 1
+    WHERE id = ? AND status = 'pending'  // optimistic lock
+    
+    if updated 0 rows → skip (otro worker lo tomó)
+    
+    try:
+      result = await executeJob(type, payload)
+      UPDATE jobs SET status = 'completed', result = ?, completed_at = now
+    catch:
+      if attempts >= max_attempts:
+        UPDATE jobs SET status = 'failed', error = ?
+      else:
+        backoff = min(1000 * 2^attempts, 60000)
+        UPDATE jobs SET status = 'pending', run_at = now + backoff, error = ?
+```
+
+## Job types
+- `exploration:run` — payload: { sessionId, config }
+- `report:generate` — payload: { reportId, format, filters }
+- `cleanup:old-data` — payload: { retentionDays }
+
+## NO usar Redis
+El job queue es SQLite-based para zero-dependency self-hosted.
+Es simple, funciona para el volumen esperado (decenas de jobs, no miles).