Use of the filter function in Workspace.prototype.undo has caused
problems with repeated undo/redo (see issue #7026), the
originally-chosen fix for which was the addition (in PR #7069) of
code to fireNow to post-filter the .undoStack_ and .redoStack_ of
any workspace that had just been involved in dispatching events.
This apparently resolved the issue but added considerable
additional complexity and made it difficult to reason about how
events are processed for undo/redo.
Instead, since this filtering typically does nothing (at least
nothing desirable), simply don't re-filter events on the undo
stack before replaying them.
bde216d120