Implement Eq, Clone and Hash for MemoryData and Evaluator

[ecstatic-morse]

In order to implement infinite loop detection while executing MIR, both the implementer of Machine (Evaluator) and its associated type (MemoryData), must implement Eq, Clone and Hash. This PR adds the required trait implementations.

It's possible that the Hash implementations need to be improved; only the env_vars field of Evaluator and the thread_local field of MemoryData are actually being hashed. Omitting fields from a Hash implementation is not incorrect, but could lead to collisions if the ignored fields are changing constantly. Perhaps I should instead derive Hash on a few more fields related to MIR validation?

[ecstatic-morse]

This depends on some of my commits in rustc being merged. I'm not sure how to express this requirement.

[RalfJung]

I wouldn't bother with the validation fields, but what about the memory content?

Also, this seems extremely costly -- are you essentially storing snapshots to see if one reoccurs? Why should that not eat up all RAM immediately?
(EDIT: Anyway that's to be discussed on the rustc side and I see you had some pretty awesome ideas there. :D )

[oli-obk]

but what about the memory content?

memory content is currently cared about on the rustc side, this is just for extra things that rustc doesn't know

[RalfJung]

Seems the rustc side of this got merged, so we should get this landed soon-ish?

[oli-obk]

Yea, but we need a new nightly before Travis will accept the PR

[RalfJung]

How is it possible that #395 is still green on Travis?

[oli-obk]

We didn't get a new nightly

[RalfJung]

Thanks :)