Abstract MPC beyond secret sharing based MPC

[eriktaubeneck]

I think it's fine to mention secret share as an example, but I would prefer that we don't encode that too much into the threat model since there are plenty of MPC protocols we could choose that don't use shares (e.g. shufflers).

Originally posted by @csharrison in #14 (comment)

[eriktaubeneck]

We also include secret shares in the threat model part (where we talk about aggregators having access to secret shares.) Maybe we want to be more specific, and then add other types with different analysis as they arise?