Investigate possible benefits of pre-generating claims? [JIRA: RIAK-3113]

[nickelization]

Some of us Basho folk had a discussion today about the Copysets paper and how it relates to Riak, and one byproduct of that conversation was the idea of pre-generating claims. I don't know how true this is, but at some point I recall hearing that claim v3 is superior to claim v2 in some ways, but is not used as the default because it can sometimes be very slow. If the actual data generated by the claim algorithm can be compressed to a small size, it might be worth running the algorithm in advance for a bunch of common cluster size/ring size/nval combinations and just baking that data into Riak.

Obviously there's a lot of unknowns here, and we'd have to do a non-trivial amount of work to test whether this would be worthwhile. Since we don't have time to work on it immediately, @macintux made the smart suggestion that we file an issue so as not to forget about this idea.