added doc for peer-stopping

src/aact/nodes/base.py

@@ -132,6 +132,19 @@ async def __aexit__(self, exc_type: Any, exc_value: Any, traceback: Any) -> None
     [^1]: Only if you know what you are doing. For example, in the `aact.nodes.record.RecordNode`, the `InputType` is
     `aact.messages.DataModel` because it can accept any type of message. But in most cases, you should specify the
     `InputType` and `OutputType` to be a specific subclass of `aact.messages.DataModel`.
+
+    ### Shutdown nodes
+
+    The default method for stopping nodes is through shutting down the subprocesses or RQ jobs. When you want to bring
+    down the dataflow, you can `Ctrl + c` to turn off the nodes gracefully. 
+
+    To shutdown a node itself, you can return from its event loop programatically when a certain condition is reached. 
+
+    #### Experimental feature: Peer-stopping
+    
+    An experimental feature is peer-stopping. A node can not only stop itself but also other nodes. To do this, send
+    a message to the channel `f"shutdown:{self.node_name}"` and the node manager will shutdown all of the nodes in the
+    current dataflow. 
     """
 
     input_channel_types: dict[str, Type[InputType]]