issue: interest in new reducer style?

First: **Love the boilerplate! Thank you very much!!**

The only thing is that I keep on having to rewrite the reducers, because the assumption of id being serialised and starting at 0/1, does not reflect any practical DB.

I would like to contribute a new way of doing reducers (which keeps the current upsides), while being able to deal with random ids.

Would there be any interest in that?