Validator rewards, slashing, and anti-dilution: last-price format

[rphmeier]

Intending validators each have some cumulative balance from themselves and nominators.

When choosing a set of N validators from the intentions list, all the validators should only be slashable based on the least-staked amount from the set. (e.g. if one validator puts down 1000 and the smallest puts down 5, then the guy with 1000 can only be slashed down to 995).

Reward structure can then be uniform across the set -- basing the reward on the lowest-staked validator in the set gives an incentive for the lowest amount to be as high as possible and thus for the spread to be as low as possible.

Treasury dilution under this scheme may be better based on the least-staked amount as well, since intuitively the highest-staked validators shouldn't be protected from dilution on unslashable funds.