Close archive writers when the delivery engine stops (closes #280)

The engine cached archive writers and never closed them at shutdown,
so after a clean stop an archive's rows could sit in its -wal while
the .db held no table. The engine's stop hook now evicts every cached
writer once its workers have returned, the same way deleting a webhook
does, so a clean stop leaves each archive as one file and a late write
is refused. If the workers do not return within the stop budget, the
writers are left open as a kill would leave them.

The README no longer says archives keep their sidecars across a clean
stop.

Model: opus-5-5

internal/delivery/engine.go

@@ -362,6 +362,15 @@ func (e *Engine) start() {
 // stop cancels the worker pool's context and waits for the pool
 // to drain, bounded by the stop hook's context: a wedged worker
 // must not hang the process past fx's stop timeout.
+//
+// Once the pool has drained it closes the archive writers, so a
+// clean stop leaves no archive -wal behind. Nothing else holds a
+// writer for long by then: the archive sweeper stops before the
+// engine, and deleting a webhook only closes one. If the pool did
+// not drain in time, the writers are left open, as a kill would
+// leave them: a worker still running may be mid-write, and
+// closing its writer would wait on that write and then fail the
+// next delivery the worker archives.
 func (e *Engine) stop(ctx context.Context) error {
 	e.log.Info("delivery engine stopping")
 
@@ -376,6 +385,8 @@ func (e *Engine) stop(ctx context.Context) error {
 		return err
 	}
 
+	e.dbTarget.evictAll()
+
 	e.log.Info("delivery engine stopped")
 
 	return nil