Shrink index semantics

[ninioArtillero]

Picked up the semantics exercise for the shrink index.

I added a PartialOrd instance for data PointSchedule, which stands for the assumption that we can compare PointSchedules within a shrink tree, but not in general.

On to another point: When we discussed this, you suggested
succ :: ShrinkIndex -> Maybe ShrinkIndex on the basis (if I understood correctly) that we might not have such an index.
My take here is that this is already taken into account by lookup, so in the general setting we are always able to get the succ and extend of an index (even though it might correspond to no PointSchedule).

Also, I was doubtful about specifying a next function defining how we could go about implementing the production of shrink indexes in terms of the abstract operations. I did it anyway.

As we refer to succ and extend elsewhere in the document, I think we should at least go with a reduced version of this.