feat: per-sink delivery, journal forwarding, and operational API

Persistence:
- one delivery state per sink, with claim, crash recovery, and index replacement
- retention for delivered and dead records
- LMDB writes moved off the event loop

Pipeline:
- one attempt per dispatch, with the next try scheduled by next_retry_at
- backpressure through max_queue_size
- reload re-reads YAML; SIGTERM drains and SIGHUP reloads

Plugins:
- journald source, enricher and filter processors, Loki sink
- built-ins register when entry points are missing

API and CLI:
- health, metrics, per-sink queue, and dead-letter replay
- startup prints the pipeline on one line

Tests:
- dispatch, plugins, metrics, CLI, and persistence coverage

README.md

@@ -17,8 +17,8 @@ Daena is a modular, extensible Linux service that:
 ## Architecture
 
 ```
-Sources → [Processors] → Pipeline → Persistence (LMDB) → Sink Dispatcher → Remote
-                                                              ↻ retry queue
+Sources → [Processors] → LMDB (one delivery per sink) → dispatchers
+                                                      ↻ per-sink retry, no head-of-line sleep
 ```
 
 Key design decisions:
@@ -67,8 +67,9 @@ asyncio.run(main())
 # Run tests
 uv run pytest tests/ -v
 
-# Start gRPC server
+# Start gRPC server. Startup prints the pipeline. --verbose logs each record.
 uv run daena-serve
+uv run daena-serve --verbose
 
 # Use the CLI (point at a running daena gRPC API)
 uv run daena status
@@ -160,14 +161,15 @@ The gRPC API runs on `127.0.0.1:8642` with the following RPCs:
 | RPC | Description |
 |-----|-------------|
 | `GetStatus` | Service status and version |
-| `GetHealth` | Health check (backend, plugins, runtime) |
+| `GetHealth` | Health, including oldest inflight age (`ok`, `degraded`, `down`) |
 | `GetConfig` | Loaded configuration as JSON |
 | `ListPlugins` | Registered sources, processors, and sinks |
-| `GetQueueState` | Queue state counts (pending, delivered, failed, dead) |
+| `GetQueueState` | Delivery counts, including a per-sink breakdown |
 | `ListPendingRecords` | Pending records with pagination |
 | `ListDeadRecords` | Dead-letter records with pagination |
 | `ListSinks` | Configured sinks |
-| `ReloadConfig` | Reload service configuration at runtime |
+| `ReloadConfig` | Re-read the config file and restart the pipeline |
+| `ReplayDead` | Requeue dead-letter deliveries |
 
 ### Reflect the gRPC service
 
@@ -182,9 +184,12 @@ grpcurl -plaintext 127.0.0.1:8642 describe daena.v1.Daena
 Daena provides a CLI that connects to the gRPC server:
 
 ```bash
-daena status       # Show service status
-daena health       # Check health
-daena queue        # Inspect queue state
+daena --target 127.0.0.1:8642 status
+daena health       # ok / degraded / down
+daena queue        # Inspect queue state, per sink
+daena records pending
+daena records dead
+daena replay       # Requeue dead-letter deliveries
 daena sinks        # List configured sinks
 daena plugins      # List registered plugins
 daena config       # Show loaded configuration
@@ -250,6 +255,21 @@ my_source = "my_sources:MySource"
 
 Same pattern applies for `Processor` and `Sink` base classes.
 
+Built-in plugins:
+
+| Kind | Name | Role |
+|------|------|------|
+| source | `dummy` | Interval test events (`count` stops after N) |
+| source | `journald` | `journalctl --follow` with an optional cursor file |
+| processor | `passthrough` | No-op |
+| processor | `enricher` | Hostname and static fields |
+| processor | `filter` | Drop or keep by field equality |
+| sink | `dummy` | Accept locally |
+| sink | `http` | JSON POST |
+| sink | `loki` | Loki push API |
+
+`SIGHUP` re-reads the config file (systemd `ExecReload`). `SIGTERM` drains and closes LMDB. Prometheus text is served at `http://127.0.0.1:8643/metrics` when metrics are enabled. Each sink has its own delivery state, and a failed attempt is scheduled with `next_retry_at` instead of blocking the dispatcher.
+
 ## Branch workflow
 
 - `main` — stable integration branch