Add basic Lock implementation

[sdboyer]

Patterned quite strongly after #4

[adg]

I suppose this is okay, for now. I worry that we're baking in some assumptions that gps makes, like the input hash for lock files, but let's see where it takes us.

[adg]

i'd go with i instead of k, as it's an index not a key

[sdboyer]

mm yes, fair

[sdboyer]

We could drop that memo prop now (though we'd still need the InputHash() method to fulfill the gps.Lock interface, ofc), but my choice to include it initially here was deliberate:

• It is quite decoupled from pretty much all the other data, so there's little complexity cost
• The Solutions returned from gps.Solve() will always have that hash digest available, so it's not like there's much extra work being done
• Some of our immediately-upcoming choices would be influenced by the use of that hashing system, so it struck me as better to have it in for easy, immediate experimentation purposes than to talk about it in the abstract
• The patterns of use it facilitates are likely to be helpful in thinking about overall system constraints, even if that's just in deciding we don't want to use it

Are there other assumptions there that concern you?

[sdboyer]

(merged, given the approval, but i'm happy to still discuss)

[adg]

Yeah, that all makes sense.

Are there other assumptions there that concern you?

Not that I can articulate right now. Let's see where this leads us.