Historical plots on group reports

[bheisler]

I can definitely see a need for historical plots showing all members of a benchmark group. This is not currently supported, but would be nice to have.

There is one unresolved design question though: since the user can run arbitrary subsets of their benchmarks, there may be gaps in the history of some benchmarks. How should this be handled? For that matter, it's not clear at the moment how to match entries in the history between different benchmarks at all.