Add support for Multiprocessing for PyLace's Engine.

[schmidmt]

Presently, lace.Engine in PyLace does not support Pickling and cannot be easily used with Multiprocessing.
A workaround for this is to save the Engine to disk with Engine.save; however, the ergonomics of this could be improved.

There are a few options here:

1. Add #[derive(Serialize,Deserialize)] to CoreEngine and implement pickle as a wrapper around bincode or the like. This is memory inefficient as the entire Engine, data and all, are copied to each Python instance, but it is the simplest method.
2. Implement an out-of-process, perhaps a server/client model, which allows for Multiprocessing without the memory inefficiency.

I'm open to discussing potential solutions, as there's no clear best path.