Adding a tag inconsistently affects subtasks

Related question that may or may not deserve its own bug. It would be trivial to make this change.

Imagine a hierarchy of tasks

- Task A
     - Task B
     - Task C

Tasks A and B already have tag @foo. If A is then dropped on @foo, should Task C be tagged with @foo?

Currently, if all tasks are untagged, and A is dropped on @foo, B and C will get the tag as well. But A already has @foo, it will not recurse into subtasks.

Well, now it _adds_ tags consistently (with one exception I'll get to), but it will not remove tags consistently--if any of the subtasks have been edited, the tag will remain (but this isn't true on a tag that hasn't been edited). Now, if a parent task contains edited subtasks, and you manually type in a tag, you'll end up applying a new tag for each letter that is entered (with results that look like @t @th @thi @this) in each of the edited subtasks (but it works fine in non-edited subtasks).

Also, this is my own personal preference, but I like the ability to apply tags to a parent task without affecting the subtasks. That's just me, though--that it works consistently is much more important than that. :)

The new subtask tagging scheme in r530 reverts the part that causes that regression. It also means that once the parent is closed, adding tags does not affect the subtasks.