Fixed two races.

The first is that the memory_region destructor would complain there is
still an outstanding allocation. This is because circular_buffer from
rust_chan wasn't refing its task, so the task was being destructed too
soon.

The second was where the program could deadlock while joining a
task. The target task would die in the time between checking whether
the task should block and then actually blocking. The fix is to use
the target task's lock.

src/rt/rust_builtin.cpp

@@ -393,11 +393,16 @@ task_yield(rust_task *task) {
 extern "C" CDECL void
 task_join(rust_task *task, rust_task *join_task) {
     // If the other task is already dying, we don't have to wait for it.
+    join_task->lock.lock();
     if (join_task->dead() == false) {
         join_task->tasks_waiting_to_join.push(task);
         task->block(join_task, "joining local task");
+        join_task->lock.unlock();
         task->yield(2);
     }
+    else {
+        join_task->lock.unlock();
+    }
 }
 
 /* Debug builtins for std.dbg. */