Free will still only makes sense when there's uncertainty about what's going on (that is resolved by your decisions), which is mostly interchangeable with there being many possible worlds (in your model), but if there are many (actual) worlds and no uncertainty (or only uncertainty independent of your decisions), free will doesn't happen.

In other words, many-worlds don't change anything about free will, in either direction. And to support free will, even a single branch must appear as a collection of possibilities that can't be ruled out.