Increase expansion parents from 8 to 16

[schomatis]

To conform with new analysis (see #827). Although straightforward we should check any potential impacts (e.g., parents cache size, replication time, change in the circuits).

[schomatis]

Ideally this should come after the changes in #791 I think (not blocking, though).

[porcuquine]

In keeping with notes on #827, I don't think we should change the value of the parameter. I think we should instead be generating more parents given the same parameter.

[dignifiedquire]

I think we should instead be generating more parents given the same parameter.

In this case we would need a second parameter, or something else, that gives us the upper bound of parents, as that is critical to have, especially in light of #791

[schomatis]

Will be fixed in #864.