Implement reward products

[brentstone]

Implement the reward product mechanism described in the F1 Fee distribution such that rewards are effectively auto-rebonded.

[brentstone]

Something to figure out:

If the rewards are auto-rebonded, then how should the voting power and total deltas be updated? Should the total and validator deltas be updated in place in accordance with the reward products? Should there be a way to ensure that the increase in stake due to rewards only affects the voting power starting at the pipeline_offset? When voting power is calculated, should we consult the reward products as well in addition to checking the total and validator deltas?

[tzemanovic]

We can probably take the same approach as slashing for rewards, which is to update the validators' total_deltas with calculated rewards at pipeline offset (eagerly) when it's calculated and individual bonds amount only when those bond are being withdrawn (lazily)

[brentstone]

Implemented in #714