feat: cola de jobs asincrona basada en SQLite (issue 0013)

Implementa el subsistema completo de background jobs para apps Go en el
dominio infra. 9 funciones + 3 tipos + 17 tests, todos pasando.

- Tipos: Job (product), JobQueue (product), JobStatus (sum) con
  JobHandler, EnqueueOption y WorkerOption usando functional options pattern
- job_queue_create: CREATE TABLE + indices + WAL mode
- job_enqueue: INSERT con UUID (github.com/google/uuid), WithPriority/WithScheduledAt/WithMaxAttempts
- job_dequeue: SELECT+UPDATE atomico en transaccion exclusiva, filtro por jobTypes
- job_complete / job_fail: transiciones de estado; fail → dead cuando attempts >= max_attempts
- job_status_summary: pura, formatea conteo de jobs por estado
- job_worker: poll loop bloqueante, context-cancelable, graceful shutdown
- job_worker_pool: N workers con golang.org/x/sync/errgroup
- job_cleanup: DELETE jobs terminales mas viejos que olderThan

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

functions/infra/job_fail.md

@@ -0,0 +1,43 @@
+---
+name: job_fail
+kind: function
+lang: go
+domain: infra
+version: "1.0.0"
+purity: impure
+signature: "func JobFail(q *JobQueue, jobID string, errMsg string) error"
+description: "Incrementa el contador de intentos y transiciona el job a 'failed' (reintentable) o 'dead' (sin mas intentos) segun si attempts >= max_attempts. Almacena el mensaje de error."
+tags: [job, queue, fail, retry, dead, sqlite, async, background, infra]
+uses_functions: []
+uses_types: [job_queue_go_infra]
+returns: []
+returns_optional: false
+error_type: "error_go_core"
+imports: [fmt, time]
+params:
+  - name: q
+    desc: "cola de jobs creada con JobQueueCreate"
+  - name: jobID
+    desc: "UUID del job que fallo"
+  - name: errMsg
+    desc: "mensaje de error para almacenar en la columna error"
+output: "error si el job no existe o falla el UPDATE"
+tested: true
+tests:
+  - "fail_increments_attempts"
+  - "fail_transitions_to_dead"
+test_file_path: "functions/infra/job_queue_test.go"
+file_path: "functions/infra/job_fail.go"
+---
+
+## Ejemplo
+
+```go
+if err := handler(job); err != nil {
+    _ = JobFail(q, job.ID, err.Error())
+}
+```
+
+## Notas
+
+El UPDATE es atomico: incrementa attempts y decide el nuevo status en un solo SQL con CASE WHEN. Si `(attempts + 1) >= max_attempts` → status='dead', sino → status='failed'. Los jobs failed pueden ser reintentados manualmente reseteando su status a 'pending'. Los jobs dead no se procesan automaticamente.