Self-review

distributed/worker_state_machine.py

@@ -564,6 +564,7 @@ class StealResponseMsg(SendMessageToScheduler):
 @dataclass
 class StateMachineEvent:
     """Base abstract class for all stimuli that can modify the worker state"""
+
     __slots__ = ("stimulus_id", "handled")
     #: Unique ID of the event
     stimulus_id: str

docs/source/scheduling-state.rst

@@ -80,10 +80,11 @@ forgotten
     dereferenced from the scheduler.
 
 .. note::
-    There's no intermediate state between waiting ``waiting`` and `no-worker``: as soon
-    as a task has all of its dependencies in memory somewhere on the cluster, it is
-    immediately assigned to a worker. This can lead to very long task queues on the
-    workers, which are then rebalanced dynamically through :doc:`work-stealing`.
+    There's no intermediate state between ``waiting`` / ``no-worker`` and
+    ``processing``: as soon as a task has all of its dependencies in memory somewhere on
+    the cluster, it is immediately assigned to a worker. This can lead to very long task
+    queues on the workers, which are then rebalanced dynamically through
+    :doc:`work-stealing`.
 
 In addition to the literal state, though, other information needs to be
 kept and updated about each task.  Individual task state is stored in an