fix(worker): claim Running before creating run resources

TaskRunner.RunAsync created the worktree (PrepareRunDirectoryAsync)
before claiming Running via StartRunningAsync. RunNow dispatches by
task id with no atomic claim of their own, so a Queued task racing
the queue picker's atomic SQL claim could hit WorktreeManager's
branch-collision self-heal, which force-removes and recreates the
winner's live worktree mid-run.

Move the claim before any resource creation and bail out immediately
when it's rejected. OverrideSlotService.RunNow also fast-rejects a
task already Running in the DB (defense in depth). RunCancellationRegistry
now refuses (and logs) a double registration instead of silently
overwriting the first CTS, so a losing dispatch's cleanup can no
longer unregister the winner's cancellation token.

tests/ClaudeDo.Worker.Tests/External/AddSubtaskToolTests.cs

@@ -74,7 +74,7 @@ public sealed class AddSubtaskToolTests : IDisposable
             NullLogger<TaskRunner>.Instance, state, new TaskRunTokenRegistry(), new AttachmentStore(), new FakeSessionSkillSeeder(), new FakeTranscriptUsageReader());
         var waker = new ClaudeDo.Worker.Queue.QueueWaker();
         var picker = new ClaudeDo.Worker.Queue.QueuePicker(dbFactory);
-        var runCancels = new RunCancellationRegistry();
+        var runCancels = new RunCancellationRegistry(NullLogger<RunCancellationRegistry>.Instance);
         var overrideSlot = new OverrideSlotService(dbFactory, runner, NullLogger<OverrideSlotService>.Instance, runCancels);
         var queue = new QueueService(dbFactory, runner, cfg, NullLogger<QueueService>.Instance, waker, picker, overrideSlot, state, runCancels,
             new FakeUsageGate(), new UsageState(), broadcaster);