Setting hashes with Mutators breaks Backbone atomicity

Consider `model.set({key1: val1, key2: val2});`

In traditional Backbone, this will set both keys to their respective values and then fire `change:key1` and `change:key2` events. In other words, Backbone guarantees the atomicity of set calls with hashes. Mutators breaks this contract by firing each change event after it's value is set. So Mutators would set key1 to val1, then fire the `change:key1` event, then set key2 to val2, then fire the `change:key2` event.

I believe this is because of the way that calls to `oldSet` works in the Mutators source.