Since this PR builds upon #97 it already has two phase key aggregation so the change is very simple. The other thing to note is that we were already "pre-negating" the secret keys of keypairs so we don't get any reduction in the number of negation flags we have to keep around (we were down to one already).
Experimenting with the idea I put forward here: https://github.com/jonasnick/bips/issues/32.
Since this PR builds upon #97 it already has two phase key aggregation so the change is very simple. The other thing to note is that we were already "pre-negating" the secret keys of keypairs so we don't get any reduction in the number of negation flags we have to keep around (we were down to one already).